nta review of the effectiveness of treatment for alcohol problems fullreport 2006 alcohol2, Study notes of United Kingdom Philosophy

Download nta review of the effectiveness of treatment for alcohol problems fullreport 2006 alcohol2 and more Study notes United Kingdom Philosophy in PDF only on Docsity! Review of the effectiveness of treatment for alcohol problems Duncan Raistrick, Nick Heather and Christine Godfrey National Treatment Agency for Substance Misuse 8th floor, Hercules House, Hercules Road, London SE1 7DU Tel 020 7261 8573 Fax 020 7261 8883 Email: nta.enquiries@nta-nhs.org.uk Website: www.nta.nhs.uk Publications All NTA publications can be downloaded from www.nta.nhs.uk. To order additional copies of this report, complete the online order form at www.nta.nhs.uk. Alternatively, email nta@prolog.uk.com or telephone 08701 555 455 and quote product code ALCOHOL2 R eview of the effectiveness of treatm ent for alcohol problem s NFS Cover with spine 1/11/06 10:36 Page 1 Foreword Foreword This timely, authoritative and comprehensive review of what research tells us about alcohol treatment is very welcome. Alcohol misuse represents a significant burden to the NHS and wider society. Both The Alcohol Harm Reduction Strategy for England (2004) and the Choosing Health White Paper (2005) identified a need for better identification and treatment of alcohol problems. The evidence base reviewed here informed the publication of Models of Care for Alcohol Misusers (2006), which provides clear guidance on the development of local systems to identify and intervene with alcohol misuse problems. This review offers practitioners, as well as commissioners and managers of services, the information they need to ensure that what they provide reflects the best available evidence. This review covers the published international research literature on alcohol interventions and treatment. In describing the effectiveness of the various interventions and treatments available it will enable local services and partnerships to assess current provision and plan future developments to meet the needs of their populations. Our relationship with alcohol as a society is complex. A source of pleasure and enjoyment for many it is also implicated in many of the most challenging problems we encounter. This review addresses the techniques for intervening early to identify excessive and risky alcohol use as well as the approaches for dealing with developed problems. UK and international research informs us that alcohol treatment can be an effective and cost effective response to alcohol problems. While there is compelling evidence for investment in alcohol treatment, this review makes clear that it will be essential to invest wisely in interventions of proven effectiveness. In order to prevent harm associated with alcohol misuse and to treat people with alcohol problems effectively, local partnerships will need to commission and deliver effective, integrated solutions. I believe this publication is a key reference tool to facilitate the development of effective local alcohol treatment systems that can contribute to reduced alcohol-related harm in our communities. I congratulate the authors on their achievement and have no hesitation in commending this review to service providers, commissioners and anyone else with an interest in alcohol treatment. Baroness Massey of Darwen Chair, National Treatment Agency for Substance Misuse 5 0 prelims.qxp 17/11/2006 11:54 Page 5 Review of the effectiveness of treatment for alcohol problems 6 Annette Dale-Perera, director of quality, National Treatment Agency Dr Emily Finch, clinical team psychiatrist, National Treatment Agency Tim Murray, policy officer, National Treatment Agency Professor Colin Drummond, professor of addiction psychiatry, St George’s, University of London Dr William Shanahan, lead clinician, Central and North West London Mental Health NHS Trust Substance Misuse and Prison Services Richard Phillips, acting chief executive, Alcohol Concern Professor IT Gilmore MD PRCP, president, Royal College of Physicians Dr Duncan Raistrick, associate medical director, Leeds Mental Health Trust Professor Nick Heather, emeritus professor of alcohol and other drug studies, Northumbria University Dr Linda Harris MRCGP, clinical director, Wakefield Integrated Substance Misuse Services and RCGP Substance Misuse Unit The steering group 0 prelims.qxp 17/11/2006 11:54 Page 6 Contents 7 Ten key themes ..........................................................9 1 The review process ..........................................13 1.1 Introduction ..............................................13 1.2 Policy context...........................................14 1.3 Objectives ................................................14 1.4 Terminology..............................................15 1.5 Chapter structure .....................................17 1.6 Summary..................................................17 2 Broadening the base of treatment and interventions.............................19 2.1 Introduction ..............................................19 2.2 Categories of alcohol misuse ...................19 2.3 Prevalence ...............................................22 2.4 Goals of treatment....................................23 2.5 Including family and friends in treatment ..................................25 2.6 Service user choice ..................................26 2.7 Increasing accessibility and responsiveness of treatment.....................26 2.8 Stepped care ...........................................27 3 Recent evidence on treatment effectiveness....................................31 3.1 Background .............................................31 3.2 Equivalence of outcomes for psychosocial treatments...........................31 3.3 The Mesa Grande project.........................32 3.4 Systematic reviews commissioned by governments .......................................34 3.5 Project MATCH ........................................35 3.6 The United Kingdom Alcohol Treatment Trial ..........................................39 3.7 Implications for treatment practice ...........41 4 Delivering better treatment ..............................47 4.1 Background .............................................47 4.2 The therapist ............................................47 4.3 Service user groups .................................50 4.4 The setting ...............................................53 5 Screening for alcohol problems ......................57 5.1 Background .............................................57 5.2 Screening questionnaires .........................57 5.3 Settings....................................................60 5.4 Biological markers ....................................61 5.5 Clinical indicators .....................................63 Appendix 1: The AUDIT Questionnaire ...............65 Appendix 2: Fast Alcohol Screening Test............66 Appendix 3: The Paddington Alcohol Test ..........67 6 Assessment and measuring treatment outcomes.........................................69 6.1 Background .............................................69 6.2 Assessment tools .....................................70 6.3 Routine follow-up .....................................75 6.4 Assessment packages .............................76 7 Brief interventions ............................................79 7.1 Background .............................................79 7.2 General effectiveness of brief interventions .....................................79 7.3 Brief interventions in primary healthcare ................................................81 7.4 Brief interventions in the general hospital ........................................82 7.5 Brief interventions in Accident and Emergency departments ...................83 7.6 Brief interventions in other medical settings ..............................84 7.7 Brief interventions in educational establishments ......................85 7.8 Brief interventions in other non-medical settings ................................86 7.9 Simple brief interventions .........................86 7.10 Extended brief interventions .....................87 7.11 Implementing brief interventions ...............89 8 Less-intensive treatment .................................93 8.1 Background .............................................93 8.2 A basic treatment scheme........................93 8.3 Condensed cognitive behavioural therapy ..................................94 8.4 Brief conjoint marital therapy ....................94 8.5 Motivational interviewing...........................95 8.6 Motivational enhancement therapy ...........98 8.7 Training in motivational interviewing ..........99 9 Alcohol-focused specialist treatment ...........103 9.1 Background ...........................................103 9.2 The community reinforcement approach................................................104 9.3 Social behaviour and network therapy .....................................105 9.4 Behavioural self-control training..............106 9.5 Behaviour contracting ............................107 9.6 Coping and social skills training..............107 9.7 Cognitive behavioural marital therapy .......................................109 9.8 Aversion therapy ....................................111 9.9 Cue exposure.........................................111 9.10 Relapse prevention.................................112 9.11 Aftercare ................................................113 9.12 Extended case monitoring......................115 Contents 0 prelims.qxp 17/11/2006 11:54 Page 7 Review of the effectiveness of treatment for alcohol problems 10 F10.- Mental and behavioural disorders due to use of alcohol F11.- Mental and behavioural disorders due to use of opioids F12.- Mental and behavioural disorders due to use of cannabinoids F13.- Mental and behavioural disorders due to use of sedatives or hypnotics F14.- Mental and behavioural disorders due to use of cocaine F15.- Mental and behavioural disorders due to use of other stimulants, including caffeine F16.- Mental and behavioural disorders due to use of hallucinogens F17.- Mental and behavioural disorders due to use of tobacco F18.- Mental and behavioural disorders due to use of volatile solvents F19.- Mental and behavioural disorders due to multiple drug use and use of other psychoactive substances F1x .0 .00 .01 .02 .03 .04 .05 .06 .07 Acute intoxication Uncomplicated With trauma or other bodily injury With other medical complications With delirium With perceptual distortions With coma With convulsions Pathological intoxication F1x .1 Harmful use F1x .2 .20 .21 .22 .23 .24 .25 .26 Dependence syndrome Currently abstinent Currently abstinent, but in a protected environment Currently on a clinically supervised maintenance or replacement regime (controlled dependence) Currently abstinent, but receiving treatment with aversive or blocking drugs Currently using the substance (active dependence) Continuous use Episodic use (dipsomania) F1x .3 .30 .31 Withdrawal state Uncomplicated With convulsions F1x .4 .40 .41 Withdrawal state with delirium Without convulsions With convulsions ICD-10 substance misuse codes 0 prelims.qxp 17/11/2006 11:54 Page 10 ICD-10 substance misuse codes 11 F1x .5 .50 .51 .52 .53 .54 .55 .56 Psychotic disorder Schizophrenia-like Predominantly delusional Predominantly hallucinatory Predominantly polymorphic Predominantly depressive symptoms Predominantly manic symptoms Mixed F1x .6 Amnesic syndrome F1x .7 .70 .71 .72 .73 .74 .75 Residual and late-onset psychotic disorder Flashbacks Personality or behaviour disorder Residual affective disorder Dementia Other persisting cognitive impairment Late-onset psychotic disorder F1x .8 Other mental and behavioural disorder F1x .9 Unspecified mental and behavioural disorder ICD-10 substance misuse codes Source: taken from ICD-10, World Health Organization, 1992 0 prelims.qxp 17/11/2006 11:54 Page 11 Review of the effectiveness of treatment for alcohol problems 12 0 prelims.qxp 17/11/2006 11:54 Page 12 The review process 15 1.4 Terminology 1.4.1 Treatment and interventions Treatment is used in the traditional sense of some specific agent, psychosocial or pharmacological, which is usually delivered by a suitably qualified individual with the intention of alleviating or resolving problems related to alcohol misuse. Treatment is something that happens within a context and it is important to understand that it is one small contributor to a much wider process of change (see chapter 15). Equally, it is important to understand that how treatment is delivered may be as important, if not more important, than what is delivered (see chapter four). Although settings may influence treatment, or may be designed as treatments in themselves, for example milieu therapy, it is generally the case that the treatments reviewed can be delivered in a variety of settings (see chapter four). Mutual aid is included as a treatment because it seems sensible to do so, on the grounds that a practitioner is not always the person delivering treatment and that many people derive great benefit from mutual aid organisations. Intervention is used as a term having a somewhat broader meaning than treatment, for example, targeted screening is an intervention rather than a treatment. Intervention includes treatment. 1.4.2 Service user This is the term most commonly used to describe people seeking help from any agency or professional. Other terms may be used when quoting directly from research. There is no particular merit attached to this description, compared to other terms such as patient, customer or client. 1.4.3 Specialist Specialist is used in the sense of a person or agency specialising in substance misuse interventions, unless otherwise stated. There are all manner of specialists, for example housing workers and liver specialists, whose specialisms are outside the substance misuse field and, therefore, are not referred to as specialists here. There are more specific uses of the term, which have been applied, for example, to different types of medical staff (see the Royal College of Psychiatry and Royal College of General Practice websites) and different levels of competency as demonstrated by a qualification (see DANOS and the Royal Colleges’ websites) but these are not intended here. 1.4.4 Diagnoses Diagnoses are those conditions recognised in the International Classification of Mental and Behavioural Disorders (ICD-10) which is widely used in the UK for statistical purposes. The ICD-10 diagnostic manual gives helpful descriptions of substance misuse and mental illness categories (World Health Organisation, 1992). ICD- 10 is thought to be more clinician-friendly than alternatives such DSM-IV, which is derived from Categories of evidence for causal relationships and treatment IA Evidence from meta-analysis of randomised controlled trials IB Evidence from at least one randomised controlled trial IIA Evidence from at least one controlled study without randomisation IIB Evidence from at least one other type of quasi-experimental study III Evidence from non-experimental descriptive studies, such as comparative studies, correlational studies and case controlled studies IV Evidence from expert committee reports or opinions and/or clinical experience of respected authorities. Proposed categories of evidence for observational relationships I Evidence from large representative population samples II Evidence from small, well-designed, but not necessarily representative samples III Evidence from non-representative surveys, case reports IV Evidence from expert committee reports or opinions and/or clinical experience of respected authorities. Table 1a: Categories of evidence Chapter 1.qxp 17/11/2006 11:04 Page 15 Review of the effectiveness of treatment for alcohol problems 16 operationally defined research criteria. DSM-IV describes individuals across five axes: i Mental illness ii Personality disorder and learning disability iii Medical conditions iv Psychosocial and environmental problems v Global assessment of functioning. Dependence is defined in ICD-10 as “a cluster of physiological, behavioural and cognitive phenomena in which the use of a substance or a class of substances takes on a much higher priority for a given individual than other behaviours”. The syndrome exists along a continuum, but it has become common practice to describe low, moderate and severe bandings. The diagnosis of dependence can be made if three or more of the following have been experienced or exhibited in the previous year: a A strong desire or sense of compulsion to take the substance b Difficulties in controlling substance use c A physiological withdrawal state d Evidence of tolerance e Progressive neglect of social activities f Continued substance use in the face of overtly harmful consequences. 1.4.5 Counselling and therapy Unless qualified by an alternative description, counselling is taken to mean client-centred or holistic therapy. Some research refers to counselling without giving a clear description of the intervention used and so counselling should not be assumed to be a precise term. Similarly, therapy is assumed to be some form of structured intervention unless qualified by an alternative description, but it is also an imprecise term. Counsellors are assumed to be qualified in counselling, except where directly reporting research studies that may not adhere to this rule. Practitioner is used as a generic term and does not imply any particular qualifications. 1.4.6 A rational approach to treatment delivery Understanding which interventions are best suited to which kinds of service user and in which settings can be difficult, so we have used a number of tools to help. In the real world, people do not fit into neat categories; nonetheless, it is useful to have a selection of models or guides to help organise thinking about treatments as, importantly, this is not about slavish adherence to a flowchart or manual. The categories of alcohol misuse, described in detail in chapter two, are intended to give an indication of the numbers of people likely to require different intensities and specialisations of treatment. In other words, this is more of a useful planning tool than a means of selecting treatment. The tiers of treatment, described in detail in MoCAM, are intended to indicate what kinds of services deliver the different intensities and specialisations of treatment – again, more useful as a commissioning tool than a means of selecting treatment. Taken together, categories of alcohol misuse and tiers of service providers are a rational way of creating and estimating the required capacity of an integrated treatment system. At a clinical level, there is no shortage of models and theories for making individual treatment decisions. We have chosen to highlight two of these. Firstly, the stepped care model, described in the next chapter, is chosen, in part, because it fits well with the main thrust of the Alcohol Harm Reduction Strategy for England (Prime Minister’s Strategy Unit, 2004). In addition, it is about intensity of treatment and maps well onto the tiers of provider and categories of alcohol misuse described in chapter two. Stepped care can be applied across agencies as well as within single providers. Secondly, the stages of change model, which is described below, is chosen in part because of its popularity and in part because it resonates with the current interest in the study of motivational treatments. Neither model has strong supportive evidence, but both have strong face validity as rational approaches. 1.4.7 Stages of change A useful tool to inform the appropriate choice of treatment is the stages of change model (Prochaska and DiClemente, 1984). The model is primarily concerned with motivation to change and the processes that lead to change. The model will be useful if placing a service user at the correct stage of change is then effective in guiding a practitioner towards the most appropriate treatment. There are four stages of change: • Pre-contemplation (including relapse) Chapter 1.qxp 17/11/2006 11:04 Page 16 The review process 17 • Contemplation (including determination) • Action • Maintenance. Pre-contemplation is characterised by a motivation to carry on drinking. People at this stage of change often use psychological mechanisms, such as rationalisation, in order to allow themselves to believe that drinking is not a problem, or to minimise the problem. The hallmark feature of the next stage, contemplation, is ambivalence or conflict – on the one hand drinking is felt to be enjoyable, or to have some utility, but on the other hand it is acknowledged to be causing problems. At the next stage, action, the conflict is removed by reaching a good-quality decision to make changes; the decision is based upon realistic expectations of how life will be better after stopping drinking or moving to problem-free drinking. Moving on from the maintenance stage of change is often the most difficult task and this stage requires continued vigilance in order to prevent relapse and a reinstatement of problem drinking. The model has been criticised on the grounds of having no sound conceptual basis, lacking evidence on the inevitability of progression through the stages and because of resistance to categorical measurement (Davidson, 1992). In contrast to this view, two versions of a Readiness to Change Questionnaire have been developed, one for the non-treatment-seeking population (Rollnick et al., 1992) and another for the treatment-seeking population (Heather et al., 1999), which can assist in assigning service users to the appropriate stage of change (Heather et al., 1993) and both are widely used. Readiness to change, measured by a different instrument, was one of the strongest predictors of outcomes in Project MATCH (Babor and Del Boca, 2003). 1.5 Chapter structure The first three chapters are concerned with setting the scene for the rest of the review. In particular, chapters two and three look at the whole range of drinkers and, in general terms, what kind of interventions are appropriate for different people. The Mesa Grande is an important plank of this review and is described in some detail in chapter three, along with recent studies that have already had, or are likely in the future to have, a high impact on practice. Chapter four is of particular importance in bringing together issues of the “how” rather than “what” of treatment; it covers the therapists who deliver treatment, the settings in which treatment may be given, and some sub-groups of help-seekers. All of these factors have an important influence on treatment outcomes. Chapters five and six are concerned with screening, assessment and measuring treatment effectiveness. Chapters 7–10 discuss the most widely used treatments available in the UK and can be considered the core of treatment. These chapters are structured by intensity and focus of the treatment. Pharmacotherapies, including detoxification, are not usually treatments on their own and are discussed in chapter 11 as enhancements to psychosocial treatment. Whether the mutual-aid movement should be considered as a treatment is debatable, but the contribution of mutual aid is immense and no review would be complete without a discussion of the subject and it is covered in chapter 12. Co-morbidity is taken in chapter 13 and is now itself the subject of a separate policy driver (Department of Health, 2002). Chapter 14, on cost-effectiveness, stands alone as having particular importance in shaping policy and, more directly, commissioning decisions at the local level. The final chapter, the treatment journey context, may be of less concern to provider agencies and of greater concern to researchers and commissioners. However, all agencies and authorities need to collaborate on working to improve alcohol treatment and this chapter is intended to be a means of helping to inform the contribution of all the different sectors in tackling alcohol problems in the UK. 1.6 Summary This review was written to support the implementation of the National Alcohol Harm Reduction Strategy and specifically to complement MoCAM. In order to avoid subjectivity, the review process took the cumulative evidence gathered by the Mesa Grande project as its starting point. We then sifted evidence of particular interest to the UK and finally cross-checked against three recent systematic reviews. Chapter 1.qxp 17/11/2006 11:04 Page 17 Review of the effectiveness of treatment for alcohol problems 20 dependence; they are not boxes in which people should be permanently placed but rough indications of current drinking patterns that individuals may move into and out of over time. Also, while it is recognised that levels of alcohol consumption, problems and dependence are imperfectly correlated with each other, these categories are intended to be pragmatically useful and to reflect the real world of service provision. In broad brush terms, different interventions are appropriate for each category of misuse (see figure 2a). 2.2.1 Hazardous drinking Hazardous drinking was described in an influential WHO report (Edwards, Arif and Hodgson, 1981) and is also termed “risky drinking”. Edwards, Arif and Hodgson (1982) defined hazardous use of a psychoactive substance as: “Use of a drug that will probably lead to harmful consequences for the user – either to dysfunction or to harm. The concept is similar to the idea of risky behaviour. For instance, smoking 20 cigarettes a day may not be accompanied by any present or actual harm but we know it is hazardous” (p7). Hazardous use of a substance is also included in the World Health Organisation’s Lexicon of Alcohol and Drug Terms (WHO, 1994), where it is defined as follows: “A pattern of substance use that increases the risk of harmful consequences for the user. Some would limit the consequences to physical and mental health (as in harmful use); some would also include social consequences. In contrast to harmful use, hazardous use refers to patterns of use that are of public health significance despite the absence of any current disorder in the individual user. The term is used currently by WHO but is not a diagnostic term in ICD-10.” This category applies to anyone drinking over recommended limits (21 units a week for men or 14 units a week for women; Royal Colleges, 1995) but without alcohol-related problems. People drinking in excess of eight units a day in men and six units a day in women (“binge drinking”) are also at increased risk of harm even although some may not exceed the “safe” weekly level. People drinking hazardously will not usually be seeking treatment for an alcohol problem, although some may realise their drinking is putting them at risk. While most will show some evidence of alcohol dependence – even if it is only an increased importance of drinking in the lifestyle – the level of dependence will be mild as measured by standard instruments; if dependence is moderate or severe, drinking is classified as “dependent”. Hazardous drinking is generally detected in primary healthcare but can also be picked up in many general hospital settings. 2.2.2 Harmful drinking Harmful drinking was also recognised in the WHO report (Edwards, Arif and Hodgson, 1981; 1982). Harmful use of a psychoactive substance is defined in ICD-10 as: “A pattern of use which is already causing damage to health. The damage may be physical or mental” (WHO, 1993). ICD-10 guidelines go on to state that harmful use should be excluded in the presence of a dependence syndrome. In this review, however, it is assumed that individuals drinking harmfully are likely to have a mild degree of dependence on alcohol, but that only moderate or severe dependence should be seen as dependent drinking per se. The harmful drinking category applies to people drinking over medically recommended levels, probably at somewhat higher levels than in hazardous drinking. However, unlike hazardous drinkers, they will show clear evidence of alcohol-related problems but often without this having resulted in their seeking treatment. The problems detected at this stage may be acute, such as an alcohol-related accident, acute pancreatitis or acute alcohol poisoning. Problems may also be of a chronic nature – for example, hypertension, cirrhosis and alcohol- related brain damage. The primary care team usually deals with these problems but they will generally also form part of the burden on the general hospital, criminal justice and social services. 2.2.3 Dependent drinking This category refers to drinking associated with an established moderate or severe level of dependence on alcohol. People who experience dependence have usually also experienced alcohol-related problems. They typically present to specialised statutory or non-statutory substance misuse services for help with the dependence itself or because of the associated health, interpersonal and social problems their dependence has caused. This group will probably be frequent attendees at general hospital services. These visits can be due to alcohol- related acute and chronic conditions and, in emergency Chapter 2.qxp 17/11/2006 11:03 Page 20 Public health programmes – primary prevention Simple brief interventions in generalist settings Severely dependent drinking Moderately dependent drinking Harmful drinking Hazardous drinking More intensive specialist treatment Less-intensive treatment in generalist or specialist settings Extended brief interventions in generalist settings Broadening the base of treatment and interventions 21 services, acute alcohol withdrawal with its range of complications including delirium tremens and alcohol withdrawal seizures at the extreme end of the spectrum. Such individuals will normally require a medically assisted detoxification, with the level of need being related to the severity of the alcohol dependence. As in the Alcohol Needs Assessment Research Project (Drummond et al., 2005), dependent drinking will be divided here into two sub-categories reflecting moderate and severe dependence. This is intended to assist service planning since these sub-categories may require quite different treatment options. 2.2.3.1 Moderately dependent drinking This sub-category applies to the majority of individuals who recognise that they have a problem with drinking, even if this recognition has only come about reluctantly through pressure from healthcare professionals, family members, employers or others. Levels of dependence are not severe and individuals have probably not reached the stage of relief drinking, that is, drinking to abolish or avoid withdrawal symptoms. However, drinkers fitting into this sub-category may experience a raised level of tolerance, symptoms of alcohol withdrawal and impaired control over drinking. In the Alcohol Needs Assessment Research None Alcohol problems Figure 2a: A spectrum of responses to alcohol problems Adapted from figure 9.1 in the Institute of Medicine [1990] report, p212. The triangle shown in figure 2a represents the population of England, with the spectrum of alcohol problems experienced by the population shown along the upper side of the figure. Responses to these problems are shown along the lower side. The dotted lines in figure 2a suggest that primary prevention, simple brief intervention, extended brief intervention and less-intensive treatment may have effects beyond their main target area. Although the figure is not drawn to scale, the prevalence in the population of each of the categories of alcohol problem is approximated by the area of the triangle occupied; most people have no alcohol problems, a very large number show risky consumption but no current problems, many have risky consumption and less serious alcohol problems, some have moderate dependence and problems and a few have severe dependence or complicated alcohol problems. Chapter 2.qxp 29/11/2006 10:04 Page 21 Review of the effectiveness of treatment for alcohol problems 22 Project they are defined as scoring 15–29 on the Severity of Alcohol Dependence Questionnaire (SADQ) (Stockwell et al., 1979). This sub-category includes a wide range of seriousness and kinds of problem. In older terminology, individuals in this category would probably not have been described as chronic alcoholics. 2.2.3.2 Severely dependent drinking This sub-category refers to the drinking of those with severe dependence and typically serious alcohol-related problems, and is the sub-category that would, in older language, have been described as applying to chronic alcoholics. Many fitting into this sub-category will have serious and longstanding problems, will typically have experienced severe alcohol withdrawal and high tolerance, and may have experienced withdrawal fits or delirium tremens; they may have formed the habit of drinking to counter or avoid incipient withdrawal symptoms. In the Alcohol Needs Assessment Research Project they are defined as scoring 30 or more on the SADQ. Many will have had several previous episodes of treatment, sometimes a large number. 2.2.4 Drinkers with complicated needs In addition to the categories above, there are groups of drinkers that may need special arrangements for treatment because of complicated needs. They include: • Those who have a co-morbid psychiatric disorder requiring more intensive support or liaison with a wider range of services such as general psychiatry (see chapter 13) • Polysubstance misusers who can present challenges in treatment due to commissioning and provision of services for either drug or alcohol misusers. People misusing drugs and alcohol may have different needs from those misusing alcohol alone and may require a different approach to treatment • Other groups that may need special consideration based on gender, age, ethnicity, disability and homelessness. It should be noted that people within these groupings are still individuals and still require an individual-focused approach (see chapter four). The scheme outlined in figure 2a provides a general indication of the kind of intervention and treatment that should normally be directed towards different categories of alcohol misuse and may be useful in planning and commissioning services. The different kinds of intervention listed here are consistent with the stepped care model of intervention and will be described in more detail later in this review. In general terms: • Primary prevention is indicated for persons drinking at low-risk levels with no alcohol problems • Simple brief interventions (simple but structured advice) in generalist settings is indicated for persons drinking hazardously with no alcohol problems but levels of consumption that put them at risk for developing such problems. Some hazardous drinkers can be offered a more extended brief intervention in the generalist setting if simple brief intervention has proved insufficient to engender change and if they are willing to accept it • While it is advisable that all alcohol misusers identified in generalist settings should be offered at least simple, structured advice, extended brief interventions in generalist settings are indicated for persons drinking harmfully who are not seeking treatment from specialist services and have not responded to simple advice. Those with relatively more serious problems and those who fail to respond to brief interventions should be persuaded to accept referral to a specialist alcohol treatment service or offered treatment in the generalist setting if resources permit • While some may respond to simple or extended brief interventions, less-intensive treatment in generalist or specialist settings is usually indicated for persons with moderate alcohol dependence who are seeking treatment. Those who fail to respond should be offered more intensive treatment • While some may respond to simple brief intervention, extended brief intervention or less-intensive treatment, more intensive treatment in specialist settings is usually indicated for people with severe alcohol dependence who are seeking treatment • Special arrangements for treatment are indicated for people with complicated needs. 2.3 Prevalence The prevalence of alcohol use disorders in the general population of England was estimated by the Alcohol Needs Assessment Research Project (Drummond et al., 2005). This based estimated prevalence across Chapter 2.qxp 17/11/2006 11:03 Page 22 Broadening the base of treatment and interventions 25 users with mild to moderate dependence (Sanchez- Craig and Lei, 1986). Even so, among drinkers with any level of dependence, a period of abstinence is advisable before moderation is attempted • Specific drinking targets should be negotiated with each service user, but moderation can be defined for treatment purposes in terms of levels of low-risk consumption recommended by medical authorities (Royal Colleges, 1995) • There are special circumstances in which the moderation goal is contra-indicated irrespective of level of dependence and where the abstinence goal should be preferred: liver damage; other medical problems that may be exacerbated by continued drinking; taking certain medications; pregnancy or an intention to become pregnant • If a service user has failed to achieve a goal of stable moderate drinking, the clinician should advise them to aim for abstinence. Conversely, if there have been failed attempts at abstinence, a moderation goal should be considered • Some service users may be thought very unlikely to be able to sustain either abstinence or moderate drinking without problems, mainly because their quality of life is so impoverished that a change in drinking offers few incentives. For these service users a harm reduction approach should be adopted in which precedence is given to modest gains in health, work and social relationships over radical changes in drinking behaviour (Heather, 1993a). For example, in the case of many homeless street drinkers, the least that can be done is to keep them as healthy as possible by occasional detoxifications and medical attention, even though an immediate return to regular excessive drinking can be expected. 2.4.2 Drinking goals among those not seeking treatment Among hazardous and harmful drinkers identified in generalist settings, the moderation goal should normally be accepted. Although a person’s preference for abstinence should always be respected, it is likely that the great majority of individuals recruited opportunistically would reject advice to abstain and would only respond to an intervention which allowed them to continue to drink, albeit at reduced levels (Heather and Robertson, 1983; Sanchez-Craig and Lei, 1986). The main advantage of including the moderation goal in treatment policy is that alcohol misusers with relatively less serious problems can be persuaded to do something about their drinking. As reflected in the Alcoholics Anonymous concept of “rock bottom” (Alcoholics Anonymous, 1939), it is often necessary for drinkers to have caused a great deal of damage to themselves, their families and to others, and to have experienced much suffering as a result, before they are prepared to consider seriously the solution of giving up alcohol for the rest of their lives. If those with less serious problems are led to believe that total and lifelong abstinence is the only solution to a drinking problem, they are likely to deny having a problem. If alcohol misusers understand that it is possible for those with less serious problems to reduce drinking to moderate levels and sustain these levels, many may find convincing reasons to try to do so. In this way, the moderation goal serves the interests of early intervention and of reducing the total aggregate of alcohol-related harm in the population at large. 2.5 Including family and friends in treatment Another sense in which a broadening of the base of treatment for alcohol problems is called for concerns the inclusion of families and friends of alcohol misusers in treatment services (Copello and Orford, 2002). This is for two principal reasons: • Family members and close friends of people with drinking problems themselves experience, or are at risk of, a range of stress-related physical and psychological disorders (West and Prinz, 1987; Moos, Finney and Cronkite, 1990) and family functioning is also adversely affected. These disorders can legitimately be called alcohol-related problems and are a proper target for alcohol treatment services (see chapters 8–10) • Evidence clearly indicates that relatives and friends can be helpful in engaging the alcohol misuser in treatment (Barber and Crisp, 1995; Miller, Meyers and Tonigan, 1999) and in bringing about a more favourable outcome of treatment (Epstein and McCrady, 1998). Methods have been developed for training relatives and friends to respond to the drinking of the alcohol misuser in ways that do not Chapter 2.qxp 17/11/2006 11:03 Page 25 Review of the effectiveness of treatment for alcohol problems 26 exacerbate the problem but are likely to assist the process of change (see chapters 8–10). On the basis of this and other evidence, Copello and Orford (2002) argue that service providers and commissioners need to consider three issues: • Models of alcohol and other drug problems should make the role played by the social environment as central and important as that played by individual factors • The base of treatment should be broadened to see the family as a legitimate unit for intervention, allowing a family member or another concerned and affected person to become the focus of help, either within a family-based intervention or as a service user in their own right • More attention and recognition should be paid to a broader set of positive outcomes from treatment in addition to reductions in alcohol use, including effects on the family and the wider social context. 2.6 Service user choice As well as choice of drinking goal, service users can also be involved in choosing the form of treatment they receive. Service user choice may be a good thing in itself but it can also improve the prospects of a successful outcome (Kissin, Platz and Su, 1970; Booth et al., 1998). This assumes that service users are provided with accurate and objective descriptions of the available options in a form they can understand. The advantages of service user choice or “self-matching” to treatment (Miller, 1989) are: • Self-matching takes place in the real world when service users seek out a form of treatment they feel they can derive benefit from and also when they fail to enter or comply with a treatment method that does not make sense to them. Given that this kind of informal self-matching occurs, it is sensible to take advantage of it and try to improve its effects • Research on human motivation generally shows that people are more likely to carry through a course of action they have chosen themselves, rather than one that has been chosen for them (Brehm and Brehm, 1981; Deci and Ryan, 1985). This freedom to choose will make it more likely that service users will comply with and complete the treatment programme, probably leading to better outcomes • More specifically, clinicians often encounter resistance to treatment from service users who deny their alcohol problems. However, resistance and denial are not so much properties of service users as characteristics of the interaction between service users and therapists (Miller and Rollnick, 2002). Service users may be less resistant to treatment and more likely to acknowledge their problems if they have played a part in choosing their own treatment and feel responsible to some degree for their progress towards recovery. Complete self-selection has been recommended (Ewing, 1977) but it is also possible to confine self-matching to a limited range of appropriate options. Service users can be involved where relevant in the following decisions: • Inpatient vs outpatient treatment setting • One-to-one vs group format • One-to-one vs with significant others • Alcohol-focused vs non-alcohol focused treatment (see chapters nine and ten) • Low vs high-intensity treatment • Motivationally based vs socially based treatment In reality, choice will be limited to situations where treatments of similar cost and effectiveness are available. 2.7 Increasing accessibility and responsiveness of treatment The 2004 Alcohol Needs Assessment Research Project (Drummond et al., 2005) showed that only a small proportion of people who might benefit from treatment for alcohol problems actually receive it. At the same time, one of the main conclusions of the present review is that there exists a range of effective treatment methods and brief interventions that can help people eliminate or reduce their alcohol problems or their risk of problems (see chapters 7–10) – hence the need to make treatment more responsive to the needs of alcohol misusers and more accessible to them. Humphreys and Tucker (2002, p127) write: “Alcohol intervention systems are often unresponsive to the full range of problems, resources, treatment preferences, Chapter 2.qxp 17/11/2006 11:03 Page 26 Broadening the base of treatment and interventions 27 goals, motivations and behaviour-change pathways within the affected population”. The extensity of treatment refers to how long treatment resources are extended over time, while its intensity refers to the amount of resources devoted to a single treatment episode. One way in which the responsiveness of treatment could be improved is by prioritising extensity over intensity in service provision. This is because: • The variation in the course of alcohol problems over time means it is a better investment to spend less healthcare resources during each contact with the service user, while allowing the intervention to extend over a longer period • The opposite and current practice of spending relatively large amounts of resources on service users for short periods is especially inappropriate for those alcohol misusers with chronic and severe problems who may need help over lengthy periods of time. A novel and inexpensive intervention of this kind is known as extended case monitoring (Stout et al., 1999) and this will be described in more detail in chapter nine. In addition to the wide dissemination of brief interventions for drinkers with less-severe problems in a range of generalist settings, there are other ways in which the accessibility and responsiveness of treatment can be increased: • Better links between the statutory and voluntary sectors • More use by healthcare professionals of mutual aid organisations (see chapter 12) • Involvement of family members and friends in facilitating entry into treatment and retention (Sisson and Azrin, 1986; O’Farrell and Cowles, 1989; Barber and Crisp, 1995; Miller, Myers and Tonigan, 1999) • Tele-health services using a range of media, including internet sites (see chapter 12) • Greater use of postal bibliotherapy programmes (see chapter 12) • Active outreach to cast a wider net in screening for hazardous or harmful drinking (see chapter five), for example in shopping centres or on the internet, and linking this screening to advice and information on helping resources of varying types and intensities • Making requirements for the receipt of services lower and more flexible • Making services more rapid and “on demand”, in order to take advantage of peaks in motivation to change. 2.8 Stepped care Stepped care refers to a way of organising services to fit with the categories of alcohol misuse described earlier in this chapter and with other aspects of the move to broaden the base of treatment. The basic principle of stepped care is that alcohol misusers are initially offered the least intrusive and least expensive intervention that is likely to be effective. Only if this first line of treatment fails is a more intensive intervention offered. If that fails, an even more intensive intervention is offered, and so on, along a scale of increasing intensity of treatment until service users show improvement (Sobell and Sobell, 2000). The stepped care model is shown in schematic form in figure 2b. In principle, the stepped care model represents a cost- effective implementation of treatment services. This is because the resources entailed in more intensive treatments are not wasted on service users who would improve with a less intensive approach. Matching service users to the intensity of treatment that fits their needs is self-selecting in the stepped care approach. Although simple in principle, there are some points to consider in the stepped care model: • The intervention and treatment modalities included in the model should be of proven effectiveness • An efficient follow-up system or some other way of monitoring progress is essential for the stepped care approach to work • Depending on the nature of their problems and the severity of dependence, service users can enter the stepped care model at any level – not necessarily the lowest point. This decision should be based on research evidence, where available, and clinical judgement • Service users should be given a substantial degree of choice over which step they enter the system at, rather than being assigned to treatment based solely on professional judgement Chapter 2.qxp 17/11/2006 11:03 Page 27 Review of the effectiveness of treatment for alcohol problems 30 Implications for… Service users and carers • Involving family and friends in treatment will improve the chances of successful treatment • There is some choice in the kinds of treatment available – the choice of drinking goal may be limited depending on the severity of problems • An abstinence drinking goal is always an option to consider. Service providers • Where possible, involve service users in choosing the setting and the general approach to treatment – choice is associated with better outcomes • Care plans will need to cover all aspects of life for the service user, not just the drinking behaviour • Clarity of drinking goal is important before starting treatment since abstinence and moderation goals call for different treatment approaches • Stepped care is a rational way of organising available resources within an agency. Commissioners • Stepped care is a rational way of organising available resources across an integrated treatment system • Interventions are required for the full range of alcohol problems, from screening for hazardous drinkers through to specialist treatment for dependent drinkers • There is an ample evidence base of clinical and cost effectiveness from which to derive commissioning plans to suit local circumstances. Researchers • Need for regular large scale surveys of the prevalence of drinking and alcohol related problems in the general population • Research to quantify the effects of user choice on outcomes • More UK research on the stepped care approach to treatment • Research into the most effective interventions for people with long-term problem drinking. Chapter 2.qxp 17/11/2006 11:03 Page 30 Recent evidence on treatment effectiveness 31 3.1 Background The purpose of this chapter is to summarise recent systematic reviews and review two large treatment trials: i The Mesa Grande project ii Other systematic reviews, including those carried out for the Health Technology Board for Scotland, for the Swedish Council on Technology Assessment in Health Care and for the Australian National Drug Strategy iii Project MATCH iv The United Kingdom Alcohol Treatment Trial (UKATT) These reviews and studies will be used to “triangulate” the conclusions of the present review. Their methods and main findings on treatment effectiveness will be briefly described in this chapter but they will be referred to at appropriate places throughout this document. The quality of treatment outcomes research has improved over the years, but many studies still have methodological deficiencies. For example, Breslin et al. (1997b) found that, regarding pre-treatment variables, only 40 per cent of studies recorded alcohol dependence, 20 per cent recorded liver function tests, and 80 per cent marital status. For treatment variables, the therapists’ training was unstated in one-third of studies and one-fifth of studies failed to describe the treatment orientation or format. Few studies use outcome measures that are not directly alcohol-related. In an attempt to take account of these deficiencies and in an effort to answer the question “what works?”, Miller et al. (2003) devised the Mesa Grande, which was taken as the basis of this review. Of the 381 studies analysed, 4.7 per cent were designed in such a way that no clear outcome could be identified and 38.3 per cent demonstrated a significant treatment effect, although this may have been judged on a single alcohol outcome and single follow-up. Similarly, meta-analyses typically depend upon one or two alcohol outcomes. In short, the treatment effectiveness literature tends to underestimate the benefits of treatment by focusing attention on drinking outcomes. 3.2 Equivalence of outcomes for psychosocial treatments In Alice in Wonderland, the Dodo Bird’s verdict was that “everybody has won, so all shall have prizes”. The phrase “dodo bird verdict” has been adopted by researchers to describe the common finding that diverse psychotherapy interventions, when compared against each other as active treatments, produce very similar outcomes (Stiles, Shapiro and Elliott, 1986). The main findings of the UKATT and Project MATCH are examples of the phenomenon – even in the case of two treatments with different theoretical underpinnings and of different intensity, there were few differences between treatment outcomes. Part of the explanation is that there are potent ingredients common to all of these therapies (Bergin and Garfield, 1994; Luborsky et al., 2002), rather than the inference that it does not matter what treatment is delivered or incorrectly concluding that treatment does not work. Moreover, because it would be unethical to set up a trial with a control group that received no treatment, trials are designed to compare a promising novel treatment against a treatment of established effectiveness (Finney, 2000). Trial designs also try to control for any variability other than in the treatments, for example therapist or site differences, that might influence the outcome. It follows that finding treatments to be equivalent is not unexpected. There are some design issues that may also contribute to the equivalence of treatments: 3.2.1 Pre-treatment motivation Motivation is thought to be a key element of behaviour change. Individuals entering similar treatment Chapter 3 Recent evidence on treatment effectiveness This third scene-setting chapter summarises the Mesa Grande Project, which has been taken as the starting point of this section and three recent systemic reviews. Two large multi-centre trials of alcohol treatment, known as Project MATCH and UKATT, are also reviewed in depth. Chapter 3.qxp 17/11/2006 11:03 Page 31 Review of the effectiveness of treatment for alcohol problems 32 programmes may have similar levels of motivation. A high proportion of individuals entering treatment, up to 20 per cent, have already achieved abstinence or started to make changes (Tober et al, 2000, p162–163; Rosengren, Downey and Donovan, 2000). It is reasonable to infer that a much higher number of help seekers are moving through the stages of change and on a trajectory towards the action stage before ever connecting with treatment services. Motivation may also be influenced by whether the treatment is offering only abstinence or moderation. 3.2.2 Therapist effects The strength of therapeutic alliance is a predictor of outcome (Babor and Del Boca, 2003, pp 55, 58) and sensitive to therapist characteristics. Therapists account for 9–40 per cent of outcome variance and are seen by some to be the essential therapeutic ingredient (see chapter four). It follows that treatment equivalence trials will attempt to control for therapist variables by attention to training of trial therapists, supervision and use of manuals. 3.2.3 Shared ingredients Different therapies have common elements. Social behaviour and network therapy (Copello et al., 2002), for example, is delivered in a motivational style, involves social network members and includes coping skills. A supportive network is a key element of 12-Step programmes and the community reinforcement approach; coping skills training may be a component of family work or a standalone treatment. Effective treatments will often have more in common than they have differences. 3.2.4 Matching The evidence on the benefits of matching service users to specific interventions is weak (Berglund, Thelander and Jonsson, 2003, p70–73). It is, however, implicit to some interventions that assessment leads to accurate selection of the most suitable treatment, as in skills training (Monti et al. 2002). Equally, some extreme characteristics might also be matched. For example, Karno et al. (2002) found people with high emotional states did best when they had the opportunity to express emotion. The more matching that takes place, the more likely that outcomes will be equivalent. 3.2.5 Post-treatment events Life events after treatment will be shaped but not determined by pre-treatment variables and the specific treatment effects. Tucker and King (1999) have suggested that the process of moving out of substance misuse evolves over several years – negative life events diminish after treatment and positive life events increase. If outcomes depend on post-treatment life events, then these are likely to occur in a similar pattern for all trial participants and, again, produce equivalent results. 3.3 The Mesa Grande project As stated in chapter one, the Mesa Grande project has been chosen as a starting point for this review. It is therefore necessary to justify this decision here. WR Miller et al. have periodically compiled systematic reviews of research on the outcome of treatment for alcohol problems. The latest of these (Miller et al., 2003) eventuated in a large table (hence Mesa Grande) in which the results of 381 trials of treatment outcome published before 2001 were summarised. Studies entering the Mesa Grande were confined to controlled trials, usually randomised controlled trials (RCTs). The great majority compared different types or intensities of treatment or the same type of treatment with and without the addition of a special therapeutic component. Controlled trials comparing at least two treatment or control conditions, and reporting post- treatment outcome on at least one measure of alcohol consumption or alcohol-related problem, were included in the review. Unpublished studies were also included if full reports describing the results were available. Two independent raters judged the methodological quality of studies on 11 dimensions, resulting in a methodological quality score (MQS) for each. Outcome logic scores (OLS) were arrived at by a similar rating process and resulted in a classification of each study as providing strong positive evidence (+2), positive evidence (+1), negative evidence (-1) or strong negative evidence (-2) for a particular treatment modality. The MQS and OLS were then multiplied for each study to arrive at a weighting of the study’s contribution to the evidence on treatment outcome by its methodological quality. These products were then summed across all studies bearing on the effectiveness of a specific treatment modality, resulting in Chapter 3.qxp 17/11/2006 11:03 Page 32 Recent evidence on treatment effectiveness 35 ii Motivational enhancement therapy iii Marital and family therapies iv Coping and social skills training. Acamprosate and supervised oral disulfiram were also recommended as adjuncts to psychosocial interventions. Given the topicality of this report and the similarity of the healthcare systems in Scotland and England, its findings are of major relevance to the present review. 3.4.2 Evidence-based review for the Swedish Council on Technology Assessment This review (Berglund, Thelander and Jonsson, 2003) is perhaps the most comprehensive synthesis of evidence on the effectiveness of treatment for alcohol and other drug problems to have appeared so far. The project was established to identify the most effective and, if possible, cost-effective interventions for alcohol and other drug problems and also those interventions already in use but not supported by research evidence. The findings of the review were intended to be used by clinicians, health administrators and policymakers to ensure the most appropriate allocation of limited healthcare resources in Sweden. With respect to treatment of alcohol problems this exercise resulted in the following general conclusions (p596): • Short-term preventive interventions by healthcare providers that target hazardous levels of alcohol consumption are shown to be effective in reducing alcohol consumption for up to two years • Many psychosocial treatment methods with a clear structure and well-defined interventions have favourable effects on alcohol problems. These methods include cognitive behavioural therapy, 12- Step treatment and structured interactional therapy strategies that involve the family in treatment • The effects of many psychosocial treatment methods (such as general counselling) have not been scientifically documented • Benzodiazepines are the most thoroughly documented medication for alcohol withdrawal. The routine practice of supplementing this treatment with anti-epileptic therapy does not have satisfactory scientific support • In long-term treatment of alcohol addiction, acamprosate and naltrexone have confirmed effects, as does disulfiram when delivered under supervision • The scientific evidence shows that treatment with antidepressants and buspirone relieves depression and anxiety in alcoholics, but it does not show any positive effects on alcohol dependence. 3.4.3 Review prepared for the National Alcohol Strategy in Australia One of the first systematic reviews of treatment for alcohol problems to include quantitative meta-analysis was carried out in Australia by Mattick and Jarvis (1993). Roughly ten years later, the Australian federal government commissioned the National Drug and Alcohol Research Centre at the University of New South Wales to update this review. An associated task was the development of updated guidelines for the treatment of alcohol problems (Shand et al., 2003b). No recommendations are given in the review document (Shand et al., 2003a), but each chapter contains one or more lists of key points emerging from the analysis contained within it. These key points and the text they summarised were consulted in the preparation of the present review. 3.5 Project MATCH One of the main reasons for conducting a meta-analysis of treatment trials is to increase sample size and statistical power. However, in the case of a well-designed trial with sufficient statistical power to detect even small effects of treatment, its findings are just as valuable as those from a meta-analysis – possibly more valuable because well- defined treatments are applied consistently across homogenous samples of service users of known characteristics and are studied under rigorous conditions. This applies to Project MATCH, which was mainly designed to investigate whether matching service users to treatments would increase the overall effectiveness of treatment. Project MATCH was the largest study of the effectiveness of treatment for alcohol problems ever mounted. The principal findings from the project were reported in Project MATCH Research Group (1997a, b; 1998a, b) and Babor and del Boca (2003) and, bearing carefully in Chapter 3.qxp 17/11/2006 11:03 Page 35 Review of the effectiveness of treatment for alcohol problems 36 mind differences in the treatments systems of the USA and England, are of major importance for this review. 3.5.1 Design and methods Project MATCH (Matching Alcoholism Treatment to Client Heterogeneity) involved nine treatment sites in the USA and a total of 1,726 clients, divided into two parallel but independent clinical trials – an outpatient arm (n=952) and an aftercare arm (n=774). The study assessed the benefits of matching clients showing alcohol dependence or abuse (DSM-III-R criteria) to three different treatments with respect to 20 client attributes. Sixteen primary and 11 secondary specific client-treatment matching hypotheses were tested. Clients within each arm of the study were randomly assigned to three 12-week, manual-guided, individually delivered interventions: • 12-Step facilitation therapy (TSF) – an approach following the principles of Alcoholics Anonymous and founded on the idea that alcoholism is a spiritual condition and a medical disease (see chapter 12) • Cognitive behavioural coping skills therapy (CBT) – an approach based on social learning theory (see chapter nine) • Motivational enhancement therapy (MET) – a less intensive form of therapy based on the principles of motivational psychology (see chapter eight). All three treatments were comprehensively laid out in manuals (Kadden et al., 1992; Miller et al., 1992; Nowinski, Baker and Carroll, 1992) and delivered by trained therapists on a one-to-one basis. CBT and TSF consisted of 12 weekly therapy sessions, while MET consisted of four sessions spread over 12 weeks. Treatment was preceded by eight hours of assessment over three sessions. There were five follow-up assessments, at post-treatment and at three-monthly intervals thereafter. The main outcome measures were the percentage of days abstinent and drinks per drinking day during the one-year post-treatment period (see Project MATCH Research Group, 1993). There was also a three- year follow-up confined to the outpatient arm (Project MATCH Research Group, 1998a). 3.5.2 Findings Matching effects: The overall objective of Project MATCH was to determine whether the careful matching of particular characteristics of clients to different forms of treatment would result in a significant improvement to the effectiveness of treatment for alcohol problems in general. This general matching hypothesis was not confirmed. Despite the general failure to find an overall improvement in treatment effectiveness through matching, the project did discover a few matching effects that can be applied in treatment programmes. These were as follows: 3.5.2.1 Psychiatric severity In the outpatient arm, clients who were low in psychiatric severity at the beginning of the trial (i.e. those with low psychiatric co-morbidity) reported more days abstinence after TSF than after CBT. This advantage for TSF had disappeared by the time of the three-year follow-up and this matching effect was not present at all in the aftercare arm. Stout et al. (2003) examined the clinical significance of this matching effect by comparing clients who were correctly matched according to the matching principle with those who were mismatched (i.e. clients with high psychiatric severity at baseline were considered matched when randomly assigned to CBT and mismatched when assigned to TSF and conversely for those with low psychiatric disturbance). They found that one year after the start of treatment, matched clients had a roughly five per cent better success rate than those who were mismatched, suggesting that only a minority of clients would benefit from the matching principle in question. 3.5.2.2 Network support for drinking In the outpatient arm only, those individuals with a social network supportive of drinking (i.e. those with numerous heavy drinking friends) did better with TSF than MET. This effect did not emerge until the three-year follow-up, implying a lag in time for the behavioural changes in question to become apparent, but when it did emerge it was the largest matching effect identified in the trial. The implication here is that clients with social networks supportive of drinking will benefit especially from a programme that encourages attendance at AA meetings, because it is the most effective means of eliminating heavy drinking friends and acquaintances from the social Chapter 3.qxp 17/11/2006 11:03 Page 36 Recent evidence on treatment effectiveness 37 network. The alternative source of (non-drinking) social support provided by the fellowship would probably be an additional factor (Connors, Tonigan and Miller, 2001). There was clear support for the hypothesised causal chain underlying this matching effect, involving degree of AA participation as a variable mediating the effect (Stout et al. 2003). As with the psychiatric severity matching effect, however, the clinical implications of the network support for drinking match were relatively modest, with clients correctly matched having a seven per cent better success rate at the three-year follow-up point than those mismatched and a three per cent better success rate than those unmatched (Stout et al., 2003). 3.5.2.3 Client anger Also specific to the outpatient arm, the finding here was that clients initially high in anger reported more days of abstinence and fewer drinks per drinking day if they had received MET than if they had received CBT. This effect persisted from the one-year to the three-year follow-up point. This finding makes sense in terms of the deliberately non- confrontational nature of MET (see chapter eight) and high client anger at initial assessment is clearly a positive indicator for the offer of MET. When clients correctly matched by the matching rule (i.e. those high in anger allocated to MET and those low in anger allocated to CBT) were compared with those mismatched, the former had a roughly ten per cent better success rate at the one-year follow-up point than the latter and a five per cent better success rate than those who were unmatched (i.e. allocated to TSF). While not a radical improvement to success rates, this superior outcome suggests that clients in outpatient programmes who are initially high in anger would be likely to benefit from being offered MET. 3.5.2.4 Alcohol dependence The only statistically significant matching effect to appear from the aftercare arm of the study was that clients low in alcohol dependence at intake reported more days abstinence with CBT than with TSF at one-year follow-up, whereas those high in dependence reported more abstinent days with TSF than with CBT. Since clients in the aftercare arm were not followed up at three years post-treatment, it is not possible to say whether this effect was a lasting one. This finding can be explained by the fact that TSF places more emphasis on total abstinence than CBT and that abstinence becomes more necessary to recovery as dependence increases (see chapter two). It also suggests that, following inpatient detoxification or day care, individuals with severe levels of dependence should be offered a 12-Step programme and those with lower dependence should be offered cognitive behavioural therapy. Project MATCH findings have no bearing on the outcome of clients in moderation-oriented programmes since, although abstinence may have been urged with different degrees of emphasis in the three treatments, moderation was never an explicit goal for any of the treatments studied. In terms of clinical effectiveness, clients matched on the principle in question had a ten per cent better outcome than those mismatched in the period 6–12 months after the beginning of treatment and a five per cent better success rate than those who were unmatched (i.e. allocated to MET) (Randall et al., 2003). 3.5.3 Main effects of treatment Although the main effects of treatment were not the intended focus of Project MATCH, they are nevertheless of considerable interest. Overall, the study showed that there were no clinically meaningful differences in success rates among the three treatments studied. This basic finding has two important aspects: 1 The effectiveness of 12-Step facilitation programmes was clearly supported. Project MATCH represented the first time a treatment programme based on 12-Step principles had been compared in a randomised trial with other commonly used and scientifically based treatments among the average run of people attending for specialist treatment for alcohol problems. As noted above, TSF was equivalent in effectiveness to the other two treatments. It must be stressed that TSF is not the same as attendance at Alcoholics Anonymous. Although it was usually delivered by “recovering alcoholics”, TSF was run on an individual basis and did not include many of the important features of AA group meetings and sponsorship. As its name suggests, TSF was intended to facilitate attendance at AA. However, this aim appears to have been successful since clients who Chapter 3.qxp 17/11/2006 11:03 Page 37 Review of the effectiveness of treatment for alcohol problems 40 3.6.1 Hypotheses UKATT hypotheses were formally expressed as null hypotheses on methodological grounds (see UKATT Research Team, 2001) but it will be more meaningful here to describe them as having a specific direction. There were two main hypotheses: 1 More intensive, socially based treatment (SBNT) will be more effective than less intensive, motivationally based treatment (MET) 2 Less intensive, motivationally based treatment (MET) will be more cost-effective than more intensive, socially based treatment (SBNT). There were also five subsidiary hypotheses involving predictions of interactions between client characteristics and treatment outcomes (matching hypotheses). These were based partly, but not completely, on client-treatment matches that had been discovered in Project MATCH. At the time of writing, the data relevant to these subsidiary hypotheses is still being analysed and will not be commented upon further. 3.6.2 Design characteristics Details of the trial design, procedures and assessments can be found in UKATT Research Team (2001). It is more relevant here to focus on some general principles and characteristics that determined the kind of trial carried out: • A pragmatic trial. In a pragmatic trial, treatments are compared under the conditions in which they would be applied in practice and the findings of the study are intended to be directly applicable to decision-making in clinical practice • An effectiveness trial. Effectiveness trials are conducted in “real world” conditions and seek to maximise external validity (generalisation to practical clinical situations) • Training, supervision and quality control of treatment delivery. In this aspect of the trial, the UKATT investigators built on the high standards set in Project MATCH (Tober et al., 2006) • Treatment process. In addition to a comparison of outcomes between two forms of treatment for alcohol problems, there was also a focus on examining treatment process (the “how” of treatment – see chapter four) by both quantitative and qualitative methods (Orford et al., 2006) • Economic evaluation. While most published studies have used retrospective data to investigate the cost- effectiveness of treating alcohol problems, in UKATT, data from clinical sites and clients was gathered concurrently with all other data, the main aim being to compare the additional costs and benefits of SBNT compared with MET and to comment on the cost- benefits applying to UKATT treatments as a whole (see chapter 14). 3.6.3 Findings Figures 3a and 3b show changes from baseline to one- year follow-up on the two main outcome measures of alcohol consumption used in the trial – percentage days of abstinence (PDA) and drinks per drinking day (DDD). The main outcomes from the trial are described in more detail by the UKATT Research Team (2005a). On each of the outcome measures in figures 3a and 3b, both groups showed marked (and statistically significant) improvements at three-month follow-up and one-year follow-up. However, there were no significant differences between groups in changes on either of these measures. The same pattern of results was seen for alcohol dependence (Leeds Dependence Questionnaire: Raistrick et al., 1994), alcohol-related problems (Alcohol Problems Questionnaire: Drummond, 1990) and psychiatric co- morbidity (General Health Questionnaire: Goldberg, 1972). To summarise, no statistically significant differences on changes in outcomes measures were observed and the first hypothesis (section 3.6.1) was therefore not confirmed. To convey better the clinical significance of UKATT findings, figure 3c shows one-year outcomes according to a classification scheme developed by Heather and Tebbutt (1989). This focuses primarily on changes in alcohol-related problems from baseline to follow-up. As will be obvious from figure 3c, there were no significant differences between groups in proportions of clients allocated to these categories. It should be noted from Figure 3c that: • Over one-quarter of clients showed a successful outcome with no alcohol-related problems at follow-up Chapter 3.qxp 17/11/2006 11:03 Page 40 Recent evidence on treatment effectiveness 41 • Forty per cent were at least much improved with a reduction in alcohol-related problems of two-thirds or more • Fifty-eight per cent were at least somewhat improved with a reduction in alcohol-related problems of one- third or more. Both UKATT treatments produced statistically significant improvements in alcohol consumption, alcohol dependence, alcohol-related problems and aspects of general functioning. It is extremely unlikely that such changes would have occurred as a result of natural recovery processes. UKATT has therefore confirmed the effectiveness of MET and found that a novel treatment, SBNT, is no less effective than MET (UKATT Research Team, 2005a). A detailed summary of UKATT findings on cost- effectiveness will be given in chapter 14. Suffice it to say here that, as might be expected in view of their differences in intensity, MET was shown to be significantly cheaper to deliver than SBNT. However, in a full societal economic evaluation, based on estimates of resources used by clients before and after treatment in the healthcare, social services and criminal justice sectors, there were no statistically significant differences between the two treatments in cost-effectiveness. The second hypothesis (section 3.6.1) was therefore not confirmed. 3.7 Implications for treatment practice Implications for treatment practice from the results so far available from UKATT will be considered in conjunction with the findings from Project MATCH. Two large multi- centre trials of treatment for alcohol problems, one in the UK and one in the USA, have now failed to find statistically significant differences in outcomes between a total of four treatment modalities that are either widely practiced or have firm foundations in theory and research. The findings of MATCH and UKATT taken with the systematic reviews are consistent with the conclusion that there is “a wealth of alternatives” (Miller et al., 1998) available for treatment in specialist services. This does not mean that all treatment methods are effective, as shown by the Mesa Grande (see page 44), or that it does not matter what treatment is given; rather, it means that there is a range of effective treatments with little research evidence of clear differences in effectiveness between them. At the present state of our research knowledge, therefore, there is no “best” treatment for alcohol Figure 3a: Mean (SD) for percentage days abstinent (PDA) from the UK Alcohol Treatment Trial Figure 3c: Categorical treatment outcomes from the UK Alcohol Treatment Trial 0 10 20 30 40 50 60 Randomised Group P D A PDA Baseline PDA Month 3 PDA Month 12 MET SBNT Figure 3b: Mean (SD) for drinks per drinking day (DDD) from the UK Alcohol Treatment Trial 0 5 10 15 20 25 30 Randomised Group D D D DDD Baseline DDD Month 3 DDD Month 12 MET SBNT 15.5 10.0 40.6 18.0 8.8 11.7 44.2 15.5 18.5 16.7 0 5 10 15 20 25 30 35 40 45 50 Abstinent Non problem drinking Much improved Somewhat improved Same/worse MET SBNT Chapter 3.qxp 17/11/2006 11:03 Page 41 Review of the effectiveness of treatment for alcohol problems 42 problems or “treatment of choice”, but a number of effective treatments that are known to be of potential benefit to clients. There is an apparent discrepancy in this chapter between the contents of the Mesa Grande, in which treatment modalities are ordered by the amount of evidence supporting their effectiveness, and the findings of Project MATCH and UKATT which failed to report clear, significant differences between a set of prominent treatments. One way of resolving this is to recall that the Mesa Grande does not directly address the comparative effectiveness of treatments but only the comparative weight of research evidence that is relevant to their effectiveness. It may be that the findings of Project MATCH and UKATT provide a truer picture by confirming the “equivalence of outcomes” but we cannot know this for certain. On the other hand, these two RCTs, however large and rigorously designed they may have been, were only two pieces of evidence compared with the 381 controlled trials included in the Mesa Grande and so may only give us a partial view of treatment effectiveness. The most reasonable conclusion here is that the apparent discrepancy in question highlights an area of uncertainty in the science of alcohol treatment: are treatments made effective by the inclusion of specific methods of behaviour change or is it non- specific factors and the way treatment is delivered, common to a range of ostensibly different treatments, that mainly account for their successful outcomes? This question is a vital one for future research but cannot be answered in this review. As noted, Project MATCH failed to discover many clinically significant matches between clients and showed that client-treatment matching, at least of the kind studied in the project, was unlikely to produce a clear, overall improvement to the effectiveness of treatment for alcohol problems in general. Nevertheless, a few client-treatment matches were discovered and these have some clinical usefulness (see chapters nine and 12). In UKATT, the investigation of such matching effects is not complete and no findings in this area are yet available. However, if it transpires that none or few indications of which types of client are suited either to MET or SBNT become apparent, the selection of treatments in practice must be made on other grounds than research evidence. These are: • Service user preference • Clinical judgement in the individual case • Existing pools of therapist training and enthusiasm for one or other treatments • Logistical considerations. One other implication for practice emerges from the findings of Project MATCH and UKATT. This is that MET, a briefer and less expensive treatment, has been shown to be as effective on the whole as three more intensive treatment modalities, CBT, TSF and SBNT, quite apart from evidence of its effectiveness from other studies. The practical implication of this is that, unless there are good grounds to offer service users more intensive treatments as a first resort, MET should be considered as the initial step in a stepped care programme within a specialist agency (see chapter two). This implication is strengthened by the fact that motivational interviewing skills, the basis upon which MET is efficiently carried out, are being increasingly taught among treatment personnel in the UK. This suggestion will be returned to in chapter eight. Chapter 3.qxp 17/11/2006 11:03 Page 42 45 Modalities with two or fewer studies Treatment modality CES N %+ Mean MQS Mean severity % Excellent Dopamine antagonist 40 2 100 10 4.00 0 Sensory deprivation 40 2 100 10 1.00 0 Biofeedback 36 2 100 13 4.00 50 Cue exposure 32 2 100 10 4.00 0 Assessment feedback (Alone) 32 2 100 8 1.00 50 Developmental counselling 28 1 100 14 2.00 100 Detoxification (alone) 26 1 100 13 4.00 0 Anticonvulsant medication 26 1 100 13 4.00 0 Treatment of significant other 26 1 100 13 3.00 0 Transcendental meditation 24 1 100 12 4.00 0 Correspondence 22 1 100 11 3.00 0 Hypnotic medication 22 1 100 11 4.00 0 Interferon 22 1 100 11 4.00 0 Contingency management 20 1 100 10 4.00 0 Affective contra-attribution 18 1 100 9 4.00 0 Tobacco cessation 14 2 50 8 3.50 0 Systematic desensitisation 13 2 50 11.5 4.00 0 Reminiscence therapy 10 1 100 10 4.00 0 Therapeutic community -4 1 0 4 3.00 0 Assessment as treatment -6 2 50 12.5 2.00 50 Moral reconation therapy -7 1 0 7 2.00 0 Apomorphine -8 1 100 8 3.00 0 Job-finding -9 1 0 9 4.00 0 Legal counselling -9 2 50 12 2.00 0 Medical monitoring -9 1 0 9 2.00 0 Minnesota model -11 1 0 11 4.00 100 Occupational therapy -11 1 0 11 3.00 0 BAC surveillance -11 1 0 11 3.00 0 Neurotherapy -12 1 0 12 4.00 0 Angiotensin-converting enzyme inhibitor -14 1 0 7 2.00 0 Choice among options -14 1 0 14 2.00 0 Buddy system -16 2 0 8 3.50 0 Dopamine agonist -16 1 0 8 3.00 0 Dopamine precursor -16 1 0 8 4.00 0 Serotonin precursor -16 1 0 8 4.00 0 Stimulant -18 1 0 9 3.00 0 Recreational therapy -22 2 0 11 4.00 0 Electrical stimulation of the head -22 1 0 11 3.00 0 BAC discrimination training -24 2 0 12 3.50 0 Beta blocker -26 1 0 13 4.00 0 Anti-psychotic medication -36 2 0 9 3.50 0 Recent evidence on treatment effectiveness Notes CES = Cumulative evidence score N = Total number of studies evaluating this modality %+ = Percentage of studies with positive finding for this modality Mean MQS = Average methodological quality score (0–17) of studies Mean severity = Average severity rating (1-4) of treated populations % Excellent = Percentage of studies with MQS >14 Reproduced with permission from Table 3 in Miller WR, Wilbourne PL and Hettema JE (2003). What works? A summary of alcohol treatment outcome research, in: Hester, R. K. and Miller, W. R. (Eds.) Handbook of Alcoholism Treatment Approaches: Effective Alternatives, p13-63 (Boston MA, Allyn and Bacon). Chapter 3.qxp 17/11/2006 11:03 Page 45 Review of the effectiveness of treatment for alcohol problems 46 Chapter 3.qxp 17/11/2006 11:03 Page 46 Delivering better treatment 47 4.1 Background Research evidence and clinical audit have exposed the variability of treatment outcomes achieved, even for essentially physical treatments, showing that outcomes frequently differ markedly from one practitioner to another and from one centre to another. It is, therefore, to be expected that for conditions such as alcohol dependence, where behaviour change is the target of treatment, specific treatment effects will be modified by other, sometimes more potent, variables: • The way treatment is delivered – therapist effects • Ethnocultural factors – particular service user groups • The place that treatment is delivered – the setting. It is a consistent finding that psychosocial treatments for problem drinkers deliver very similar results (see chapter three). Problem drinking is a context-dependent condition, that is to say that influences such as cultural norms, social networks, the regulatory system, and per capita alcohol consumption (see chapter 15) – in other words factors other than treatment – have a powerful effect on outcomes. In a meta-analysis of 72 studies, Hettema, Steele and Miller (2005) found that the effect size for ostensibly the same treatment, motivational interviewing, varied from 0–3, meaning different sites and different populations achieved very different outcomes. This chapter is concerned with some of the more important of these factors. Some of them are within the control of agencies, for example therapist competence, while some are beyond an agency’s control, for example service user characteristics. Others, for example treatment settings, may or may not be amenable to selection by service users or practitioners. 4.2 The therapist 4.2.1 Context There is an accumulation of evidence from psychotherapy showing that some therapists achieve better results than others. More effective therapists are characterised as empathic, supportive, goal-directed, helping and understanding, encouraging service user autonomy, and effective at using external resources. Less effective therapists are characterised as psychologically distant, overwhelming, belittling and blaming, intrusive and controlling, avoiding difficult issues, and self-interested (Najavits and Weiss, 1994). Meta-analyses have found that around nine per cent of the outcome variance across treatment effectiveness studies is accounted for by therapist characteristics, although in particular cases this figure may rise to between 40 and 50 per cent (Crits- Christoph and Mintz, 1991). Messer and Wampold (2002) take a more radical position and suggest that meta- analyses demonstrating treatment equivalence are best explained by common therapist characteristics, which are more powerful than the specific treatment. Trials focused on treatment effectiveness are designed to control for therapist effects (Carroll, 2001), as was the case in Project MATCH and the UK Alcohol Treatment Trial. In these circumstances, most therapists should perform within a relatively narrow range and it will not be possible to say much about the influence of therapist characteristics. It has been suggested that there may be greater variation in the performance of therapists working in substance misuse, because therapy is likely to be disrupted by service users attending while intoxicated, preoccupied with social crises or involved with the criminal justice system. There is little evidence either way. Chapter 4 Delivering better treatment This chapter is the last one before we evaluate specific treatments and is the first where we estimate the strength of evidence. The main issue here is how to deliver treatment, rather than what to deliver. The main topics covered are therapist characteristics, service user groups and settings in which to deliver services. Chapter 4.qxp 17/11/2006 11:01 Page 47 Review of the effectiveness of treatment for alcohol problems 50 4.3 Service user groups 4.3.1 Context Everyone attending a treatment service has the right to expect that their culture, gender and practical needs will be sensitively accommodated in so far as this is reasonably possible. The idea that ethno-culturally competent treatment providers (Straussner, 2001) should be able to work with all service users has appeal in that it offers both service user choice and makes best use of limited resources. However, there may be instances where local areas need to provide particular services, or elements thereof, which specifically attract, retain or provide for culturally diverse groups. Equally, it is worth searching for imaginative ways of delivering mainstream services that people from ethnic minorities wish to attend. A study in California (Weisner et al., 2002) looked at the odds ratio (how much more likely than the population as a whole) of different population groups getting into treatment. The findings were black ethnicity, 2.98; older age, 4.67; less education, 1.81; legal pressure, 7.46; work pressure, 3.57; psychiatric morbidity, 4.03. The UK would probably be different, but the point to make is that people’s lives are too complex to align them with a single special population service; perverse inclusion and exclusion criteria can quickly appear and then detract from the usefulness of a service which was set up with good intentions. Most people seeking help for a drinking problem will have certain general or common identities as well as one or more special identities. The potential for special identities is vast and may focus on any or all of demographic, social, political and other factors, including: • Gender • Sexual orientation • Professional group • Sharing a common co-morbidity diagnosis • Homelessness • Age • Ethnicity • Religion • Legal status. The list is not exhaustive and evidence is available for only a few of the groups mentioned. 4.3.2 Black and minority ethnic groups The particular rationales for speciality services for ethnic or religious groups are several: i The possibility of communicating in the service user’s first language ii The recognition and acceptance of drinking patterns that are different to the dominant culture iii The need to understand cultural or religious mores that define the relationship between service users and therapists. A detailed investigation into the key question ”Do culturally specific treatment programmes enhance the probability of successful outcome for their target populations?“ was published in Broadening the Base of Treatment for Alcohol Problems (Institute of Medicine, 1990, p356–380, 399–405). There were insufficient research findings to inform any recommendation on whether to develop services specifically for minority groups. It was recognised, however, that mainstream services would necessarily continue to be major providers for ethnic minorities and it was recommended that staff in these agencies be trained in the skills and sensitivity needed to identify and work with all minority groups. It was also recommended that minority group treatment programmes should be funded where these would improve access to treatment and where there could be proper evaluation of the service. Collins (1996) has argued that ethnic groupings are essentially a political construct with little utility in either substance use research or clinical practice. She asserts that greater variance can be found within ethnic groups than between different ethnic groups sharing, for example, a heavy drinking ethos. She suggests that ethnicity has been elevated in importance at the expense of other dimensions such as socio-economic status, education level, employment status and health. The degree of acculturation and assimilation to the majority culture is important in that the ethnicity label given to an individual may not reflect that individual’s choice of identity. There are also a very large number of groups within each major category. It is generally held that there is a low prevalence of substance misuse among ethnic minorities and the most important reason given is religious belief, but this proposition becomes less true as religious involvement is Chapter 4.qxp 17/11/2006 11:01 Page 50 Delivering better treatment 51 weakened. Karlsen et al. (1998) found a hierarchy of substance misuse among adolescents from whites (the heaviest users) to black Caribbeans, to black Africans, and to Bangladeshis. The authors found an inverse relationship of family involvement and religious influence with substance use. Among Israeli Jews, Aharonovich et al. (2001) found that the less religious, wealthier, European Ashkenazim drank more heavily than the North African and Middle Eastern Sephardim. In a study comparing perceived risks from substance use, Ma and Shive (2000) found that whites were less likely to identify risks as compared to blacks and Hispanics. In contrast, Mather and Marjot (1989) found that Asian men had twice the incidence of admissions for alcohol-related problems compared to European men – the Asian men were mostly Sikhs and Hindus. Among pregnant women, Waterson and Murray-Lyon (1989) found that 90 per cent of Europeans, 75 per cent of Afro-Caribbeans, 56 per cent of Orientals and 47 per cent of Asians were heavy drinkers before pregnancy. Orford, Johnson and Purser (2004) surveyed 1,684 individuals from second or subsequent generation black and Asian communities and found marked ethnic and gender differences in drinking; black men and women and Sikh men had patterns similar to the general population. Primary care was endorsed as a source of help whereas there was some uncertainty about the confidentiality within communities if used as a source of help. Cameron et al. (2002) speculate that the family network may make spontaneous recovery more likely among ethnic minorities – in a study of 20 Asian problem drinkers, who had “spontaneously” recovered, family honour and religious re-affiliation were frequently cited as reasons for stopping drinking. Help-seeking is strongly influenced by the experience of psychosocial problems, particularly if these are interpersonal, and by encouragement to enter treatment (see Tucker and King, 1999). Kahn et al. (2000) interviewed 31 ethnic minority drug users and 12 ethnic minority helpers about the problems of accessing services. The majority of problems related to racial origins and included the need to conceal substance use from parents and family, being reported to their parents if seen at a treatment agency, fear of unusual and severe punishments if caught, and avoiding the intolerance of the minority community. There were mixed views regarding the ideal drugs worker. The Asian community felt the need for drugs workers of the same cultural background most strongly. Hettema, Steele and Miller (2005) found that the effects of motivational interviewing were greater for ethnic minorities than whites: 0.79 against 0.26. The meta-analysis does not specify therapist characteristics. 4.3.3 Young people Services for adolescents and young people are now commissioned separately from those for adults and will have separate Models of Care guidance. The evidence suggests that the same kinds of treatment are effective for both adults and younger people (Tevyaw and Monti, 2004), but it is the social needs of young people that are often different to adults. There is a long history of health services, social care and the criminal justice system seeing young people as different from adults and in need of their own services. Young people with drinking problems tend to fall into one of two groups: those whose problems are largely related to intoxication and those whose drinking is better interpreted as a symptom of profound psychosocial disturbance (see chapter 13). It is beyond the scope of this review to elaborate on the complexities of definition, patterns of use and psychological development that are relevant to young person services. The trend towards outreach work and peer counselling has heuristic value. The report of the Health Advisory Service, The Substance of Young Needs Review 2001 (2001), gives comprehensive guidance on planning services, albeit with a focus on illicit substances. 4.3.4 Women Women have different substance using careers to men – generally they start later and respond better to treatment. Women are also more likely to bring higher rates of physical and psychiatric co-morbidity, which may complicate treatment (Davis et al., 2002). In an eight-year follow-up, Timko et al. (2002) found outcomes for women were somewhat better than for men using the same services. Similarly, a review by Jarvis (1992) concluded there are only small differences across a variety of treatment modalities and settings in the effectiveness of treatment for women compared to men but, notably, women are likely to do less well in mixed sex group therapy because of the unfavourable sexual dynamics. Furthermore, women who have been abused tend to prefer a female therapist but women who have not identified themselves as having experienced violence from men do equally well with male or female therapists (Connors et al., 1997). It is known that women differ Chapter 4.qxp 17/11/2006 11:01 Page 51 Review of the effectiveness of treatment for alcohol problems 52 significantly from men in the way that they handle the metabolism of alcohol – women are more vulnerable to organ damage, notably liver disease and brain damage, which has been attributed to having a lower volume of body fluid in which to distribute alcohol and having less first-pass metabolism, thereby causing higher blood alcohol concentrations than in males drinking similar amounts (Lieber, 2001, p.90). It is unlikely that these physiological gender differences will have any significant impact on treatment approach. 4.3.5 Homeless people Farrell et al. (1998) present data from a national survey of homeless people comprising 1,061 individuals. They note the significant association between social deprivation, psychological morbidity and substance misuse (see table 4a). Notwithstanding the mixed responses to treatment, there is a case for ensuring that the treatment system provides the basics of shelter, food and companionship for homeless people. Homeless people are a group for whom providing a special service is logical. There is evidence to support the need for a national network of services, typically residential and non-hospital, as a safety net and pathway to long-term rehabilitation. However, there has been a move away from services for homeless problem drinkers to more holistic services for the homeless. Primary care services specifically for the homeless are an example of how general medical care should now be delivered to this group through a speciality team working out of a mainstream primary care trust facility, from where help with substance misuse problems can also be provided (Wright, 2004 pp.88-102). Berglund, Thelander and Jonsson (2003) reviewed 11 randomised treatment studies of homeless people which totalled 2,527 individuals. The studies were characterised by high attrition rates but there were positive effects for behavioural treatments and for case management where this involved wraparound services. Cox et al. (1998) randomised to intensive case management (ICM) or a no- treatment control condition homeless people or those at risk of homelessness who also had an extensive history of alcohol misuse and treatment failures. The primary aims of ICM were to improve the financial and residential stability of service users and reduce their use of alcohol. At follow-up interviews carried out at six monthly intervals over two years, there were small but statistically significant differences favouring the ICM group in total income from public sources, nights spent in “own place” out of the previous 60 nights and days drinking out of the previous 30 days. Smith and Delaney (2001) compared a community reinforcement approach (CRA, see chapter nine) to standard treatment at a large day centre. The traditional CRA programme was modified by: • Adopting a group treatment format • Adding goal-setting and independent living skills groups • Adding a weekly community meeting as an opportunity for concerns to be voiced and for the social club activity to be decided • Offering a sizeable number of groups each week to allow for “misses”, without jeopardising treatment effectiveness • Using small incentives for attendance • Allowing interested individuals to participate even if they were unwilling or unable to take disulfiram • Providing housing for clients in both treatment and control conditions throughout the programme. Large reductions in drinking were found in both groups at one year. However, the CRA group showed consistently Weekly alcohol units Men 22+ Women 15+ Any drug use (including cannabis) Any drug use (excluding cannabis) Smoking over 20 cigarettes per day Hostel residents 22% 11% 3% 34% Private sector residents 9% 7% 1% 18% Night shelters 52% 29% 11% 43% Sleeping rough 55% 24% 6% 46% Table 4a: Substance use among homeless people using different abodes Chapter 4.qxp 17/11/2006 11:01 Page 52 Delivering better treatment 55 6 Drug treatment programmes are much better developed but not always integrated with alcohol programmes. McMurran (2005) has reviewed prison treatment programmes and found only one, which was aimed at drink drivers, accredited specifically for alcohol-related offenders. While research may be lacking there are comprehensive treatment guidelines with accompanying clinical tools (HM Prison Service and Department of Health, 2004) available for prison healthcare staff. 4.4.6 Conclusions • The evidence base for determining the optimal treatment setting is weak because treatment has usually been delivered in what has been considered the safest and, to a lesser extent, cheapest setting. Service user choice may change these considerations (IV) • There is a need to have residential treatment facilities for selected groups of service users (IIB). Chapter 4.qxp 17/11/2006 11:01 Page 55 Review of the effectiveness of treatment for alcohol problems 56 Implications for… Service users and carers • Service user groups can help agencies to be more user friendly and help to build services that are ethnoculturally competent • Expect that services are able to offer a choice of treatment settings • Expect that good treatment will be an active and participative process of working alongside a therapist • Service user groups could be given the lead on developing volunteer schemes to be active in supporting agencies. Service providers • Recognise the importance of general therapist characteristics such as attitudes and appearance - ensure that staff receive good quality training and supervision • Consider the benefits of using manual guided treatments • Ensure that, where appropriate, services can be delivered in a variety of settings such as service users’ homes • Ensure that staff receive diversity training and understand how to apply this knowledge to treatment delivery • Be open to input from service user groups on how to make services user friendly and particularly how to attract minority groups. Commissioners • There needs to be good provision for the needs of special groups within the locality – this may be achieved through generic or specialist services • There need to be imaginative ways of making access to services more user friendly and at the same time retaining the cost and flexibility benefits of larger agencies • There is a need to ensure the availability of residential facilities for defined service user groups • Contracts should include minimum standards for staff training and supervision • Wraparound services are especially valuable for some service user groups, such as the homeless. Researchers • A key issue is the relative contribution to outcomes of therapists, pre-treatment service user characteristics and specific treatments • The cost effectiveness of domicillary versus centre-based care needs investigation • Research is needed to determine which service user groups require residential care. Chapter 4.qxp 17/11/2006 11:01 Page 56 Screening for alcohol problems 57 5.1 Background Identification of alcohol misuse among people not seeking treatment for alcohol problems can be done in three ways: 1 Screening questionnaires in printed or electronic form, for service users to complete or practitioners to read out. Screening questionnaires are more likely to be answered accurately when: – The practitioner administering the instrument is friendly and non-threatening – The purpose of the questions is clearly related to the service user’s health status – If possible, the service user is alcohol- and drug - free at the time – The information is seen as confidential – The questions are easy to understand (Anderson, 1996). 2 Biological markers of recent alcohol consumption 3 Clinical indicators by clinicians using clinical history or signs at physical examination. A good screening method should have both high sensitivity and specificity: • Sensitivity is the proportion of alcohol misusers who are screened positive by the test • Specificity is the proportion of those who are not alcohol misusers who are screened negative by the test. 5.2 Screening questionnaires 5.2.1 Context General purpose screening can be carried out in non- medical settings – educational, criminal justice, social service and workplace settings. A key issue for all screening programmes is whether to target at-risk groups or the whole population. Two recent articles by Beich and colleagues (Beich, Gannik and Malterud, 2002; Beich, Thorsen and Rollnick, 2003) concluded that screening created more problems than it solved and did not seem to be an effective precursor to brief interventions targeting excessive alcohol use. The conclusions reached by Beich et al. have been strongly criticised and have led to a heated controversy (see correspondence on www.bmj.com from 18/10/2002 to 1/12/2002 and from 4/9/2003 to 7/3/2004). Targeted rather than universal screening was recommended in the Alcohol Harm Reduction Strategy for England (Prime Minister’s Strategy Unit, 2004: p42). A form of targeted screening in primary healthcare was described by Israel et al. (1996). This consisted of a trauma scale developed by Skinner et al. (1984) based on evidence of a high correlation between the occurrence of trauma and alcohol misuse. Israel et al. (1996) reported that the use of their trauma scale method identified 62–85 per cent of the expected number of alcohol misusers in a primary healthcare population. The method was acceptable to both patients and practitioners. In a survey of expert opinion on screening and brief interventions in primary healthcare in the UK (Heather et al., 2004), there was a clear consensus among experts on confining routine screening to new patient registrations, general health checks and special types of consultation. This finding was also supported by the results from focus groups among primary healthcare professionals and patients (Hutchings et al., 2006). Chapter 5 Screening for alcohol problems Before reviewing treatments themselves, in his chapter we cover the topic of screening. We review commonly used screening tools, biological markers and clinical markers of alcohol misuse. Early detection is an essential element of broadening the base of treatment to detect problem drinkers before they become help-seekers. Chapter 5.qxp 17/11/2006 11:01 Page 57 Review of the effectiveness of treatment for alcohol problems 60 hazardous drinkers with an accuracy of 97 per cent (Hodgson et al., 2002). The complete procedure had a sensitivity greater than 91 per cent and a specificity greater than 86 per cent in the four settings in which it was validated. 5.2.2.6 Michigan Alcoholism Screening Test (MAST) The MAST is a 24-item screening instrument originally described by Selzer (1971). It also comes in a 13-item shortened form (SMAST; Selzer, Vinokur and Van Rooijen, 1975) and a ten-item brief form (BMAST; Pokorney, Miller and Kaplan, 1972). It has been extensively used in research and treatment circles over the years. As its name suggests, the MAST was developed to detect severe alcohol dependence, including early signs of dependence. Its main advantage in screening is that it provides an individual’s responses to a range of possible alcohol-related problems and signs of dependence that may be useful in assessment. Its main disadvantage for screening hazardous and harmful alcohol consumption is that it asks “ever” questions, which apply to the respondent’s lifetime. This neglects the fluctuation of alcohol consumption and problems over the course of time. Evidence reviewed above shows that, although it may be as efficient for the detection of alcohol dependence, it is inferior to the AUDIT for the detection of hazardous and harmful consumption. 5.2.3 Conclusions • The AUDIT is a screening instrument of good sensitivity and specificity for detecting hazardous and harmful drinking among people not seeking treatment for alcohol problems (III) • The AUDIT is has been validated for use in a wide range of settings, populations and cultural groups and is in widespread use worldwide (II) • The AUDIT is superior to the MAST and CAGE for the detection of hazardous and harmful drinking, although not necessarily in the detection of significant alcohol dependence (II) • The AUDIT can be embedded in a general health questionnaire without loss of efficiency (III) • The AUDIT should be considered as the screening instrument of first choice in community settings • Shortened versions of the AUDIT can be used in very busy settings without undue loss of efficiency compared to the full AUDIT (III) • The AUDIT-C is based on consumption items alone and is an efficient tool for the detection of hazardous drinking (II) • The FAST offers a rapid and efficient way of screening for hazardous and harmful alcohol consumption that can be used in a variety of settings (II). 5.3 Settings 5.3.1 Antenatal clinics 5.3.1.1 Context Given the risk of harm to the unborn foetus from the mother’s excessive drinking, the detection of alcohol misuse among pregnant women is of major importance. Two screening instruments, both taking approximately one minute to complete, have been developed to screen for hazardous and harmful drinking among pregnant women: • T-ACE (Sokol, Martier and Ager, 1989) is a four-item adaptation of the CAGE • TWEAK (Russell, 1994) is five-item instrument using items from the CAGE and MAST. 5.3.1.2 Evidence Research on the efficiency of the T-ACE and TWEAK is reviewed by Dawe et al. (2002a). • The T-ACE has consistently been shown to be of higher sensitivity and specificity for detecting alcohol misuse among pregnant women that the MAST or CAGE (Russell et al., 1996; Chang et al., 1998) • The TWEAK appears to be somewhat more sensitive and less specific than the T-ACE but both are clearly more efficient than the MAST or CAGE (Russell, 1994) • The superiority of the TWEAK for screening in pregnancy has been demonstrated in a wide range of socio-economically and ethnically diverse populations in the USA (Russell et al., 1996; Chang et al., 1999a) • The TWEAK also appears to be an efficient screening tool among men and non-pregnant women (Dawe et al., 2002a). Chapter 5.qxp 17/11/2006 11:01 Page 60 Screening for alcohol problems 61 5.3.1.3 Conclusions • Both the T-ACE and TWEAK are superior screening instruments for detecting alcohol misuse among pregnant women than the MAST or CAGE (III) • The TWEAK seems to be more sensitive but less specific than the T-ACE (III) 5.3.2 A&E departments 5.3.2.1 Context Pressure on time for screening is particularly relevant to the A&E setting and there may also be special difficulties in screening among injured and intoxicated patients. Nevertheless, it is possible to screen efficiently for alcohol misuse in A&E settings and to refer those screening positive for brief interventions (Green et al., 1993; Huntly et al., 2001). The FAST (see above) was developed for use in A&E settings. Another such instrument is the RAP24 developed by Cherpitel (2000) in the USA, although the efficiency of this instrument has yet to be established. In the UK, Smith et al. (1996) described the development of the Paddington Alcohol Test (PAT), which is shown on page 67. The PAT takes less than one minute to complete and was designed for ease of administration and relevance to presenting problems in the A&E setting. To increase the prospects for implementation, the PAT comes with guidance to practitioners as to the top ten types of presentation in which it should be applied (Huntley et al., 2001). 5.3.2.2 Evidence Hodgson et al. (2003) compared the FAST with the PAT and CAGE in four UK A&E departments. All three tests were quicker to administer than the full AUDIT, with the FAST taking 12 seconds on average. All tests identified drinkers who would accept a health education booklet (over 70 per cent) or five minutes of advice (over 40 per cent). The FAST was consistently reliable when sensitivity and specificity were tested against the AUDIT as a gold standard. The FAST had better sensitivity and specificity than the PAT, though in this study an older version of the PAT was used. Using a newer version of the PAT, Patton et al. (2004) reported that the PAT showed good concordance with the full AUDIT, but could be administered in one-fifth of the time. Huntley et al. (2001) reported that the uptake of the PAT by senior house officers was improved when their performance was audited and they were given feedback on it. Rates of detection of alcohol misuse among A&E patients showed a four-fold increase as a result. The selective screening forming part of the PAT procedure was calculated to account for 77 per cent of hazardous drinkers presenting to A&E departments. An analysis of feedback to patients screening positively on the PAT showed that this increased the proportion willing to accept brief interventions by 23 per cent (Patton, Crawford and Touquet, 2003). 5.3.2.3 Conclusions • The FAST is a rapid and efficient screening tool for detecting alcohol misuse in the A&E setting (III) • The PAT has been developed to fit with the demands of very busy A&E departments and is a quick and efficient screening tool in this setting (III). 5.4 Biological markers 5.4.1 Context A possible disadvantage of screening questionnaires is that they are based on self-reports of alcohol consumption and problems and may therefore be inaccurate to varying degrees. Although self-reports are more reliable and valid than is sometimes supposed (Babor et al., 2000), they can be influenced by deliberate under- or overestimation of consumption and by failures of memory and other cognitive factors. While laboratory measures can increase confidence in the reliability of self- reports, they add little information that cannot be gained more cheaply and efficiently by self-report. Aertgeerts et al. (2001) also found laboratory measures to be far less sensitive for the detection of alcohol misuse in primary healthcare settings. Biological markers may also be used as part of a comprehensive assessment and as treatment outcome measures (see chapter six). Biological markers of alcohol consumption have the advantage that they are completely objective and cannot be distorted in the same way as questionnaires. In certain circumstances, notably legal proceedings or health checks for employees in high-risk occupations, it may be necessary to have the additional evidence of an objective measure. There are ethical issues in that investigations Chapter 5.qxp 17/11/2006 11:01 Page 61 Review of the effectiveness of treatment for alcohol problems 62 used for the purposes of detecting illness may also indicate excessive drinking, so practitioners need to ensure that service users are properly informed of the reasons for taking blood tests and the risks of later disclosure. While the search for improved markers of alcohol consumption continues (Whitfield, 2001), the following are currently used to detect levels of alcohol consumption: • Blood or breath alcohol concentration • Mean corpuscular volume (MCV) • Serum gamma-glutamyltransferase (GGT) • Aspartate aminotransferase • Alanine aminotransferase • Carbohydrate deficient transferrin (CDT) • HDL-cholesterol • Uric acid. We will consider GGT, CDT, and MCV, which are often used as markers of consumption, whereas the other investigations are more usually used to detect pathology and are incidental markers of alcohol intake. Conigrave et al. (2003) concluded that none of these markers are well suited as screening tests, but much more useful as opportunistic diagnostic tests or for monitoring change when abnormal at baseline. Direct measurement of ethanol levels can be useful. 5.4.2 Evidence 5.4.2.1 Gamma-glutamyltransferase (GGT) GGT is a liver enzyme and the most commonly used biochemical marker of alcohol consumption. Drinking four or more drinks per day for four to eight weeks significantly raises levels of GGT in alcohol dependent individuals, while four to five weeks of abstinence usually returns levels to within the normal range (Allen and Litten, 2001). GGT is raised in between 60 and 80 per cent of those severely dependent on alcohol. However, it can also be raised by non-alcoholic liver disease, certain medications and obesity, leading to false positives (i.e. poor specificity) on this test. The proportion of heavy drinkers with raised GGT is between 20 and 50 per cent (Whitfield, 2001). This makes GGT of little value for detecting hazardous and harmful drinking in community settings. The GGT, with a half-life of approximately 21 days, is reasonably sensitive to short-term changes in consumption and has been found to be a predictor of all cause mortality. Feedback of GGT was one of the principal ingredients in a pioneering study of brief interventions carried out as part of a population health screening programme in Sweden (Kristenson et al., 1983), indicating its potential usefulness as a therapeutic device. 5.4.2.2 Carbohydrate deficient transferrin (CDT) Unlike other liver enzymes, elevated values of CDT are almost entirely specific to alcohol metabolism and reflect the level of recent alcohol consumption. CDT tests have a low rate of false positives and are sensitive to moderate levels of consumption (Javors and Johnson, 2003). CDT becomes elevated earlier in response to heavy drinking than GGT (Allen et al., 2001). Laboratory analysis is relatively expensive. In a review of 54 studies comparing CDT to other laboratory markers, Salaspuro (1999) found that: • CDT was slightly more sensitive than GGT in detecting changes to drinking over a 3–4 week period • CDT was similar to GGT in detecting alcohol misuse in males • There was mixed evidence of the relative efficiency of CDT and GGT among females • CDT showed low sensitivity in detecting lower levels of hazardous drinking in community samples • CDT was superior to GGT in detecting alcohol misuse among individuals with alcohol-related and non- alcohol-related liver disease • CDT was overall marginally superior to other laboratory markers. 5.4.2.3 Mean corpuscular volume (MCV) MCV is an index of red blood cell size which increases with excessive drinking after four to eight weeks. Although more specific than other tests, MCV has very low sensitivity for the detection of heavy drinking (Helander, 2001). 5.4.2.4 Ethanol The direct measurement of ethanol levels can be achieved using a breathalyser or blood test. This may be useful both as feedback to service users and to give practitioners an indication of the service user’s tolerance, which in turn reflects the previous pattern of drinking. Urine alcohol concentration is a crude measure that may Chapter 5.qxp 17/11/2006 11:01 Page 62 Screening for alcohol problems 65 1. How often do you have a drink containing alcohol? (0) Never (1) Less than monthly (2) 2–4 times a month (3) 2–3 times a week (4) 4 or more times a week 2. How many units of alcohol do you drink on a typical day when you are drinking? (0) 1 or 2 (1) 3 or 4 (2) 5 or 6 (3) 7, 8 or 9 (4) 10 or more 3. How often do you have six or more units of alcohol on one occasion? (0) Never (1) Less than monthly (2) Monthly (3) Weekly (4) Daily or almost daily 4. How often during the last year have you found that you were not able to stop drinking once you had started? (0) Never (1) Less than monthly (2) Monthly (3) Weekly (4) Daily or almost daily 5. How often during the last year have you failed to do what was normally expected from you because of drinking? (0) Never (1) Less than monthly (2) Monthly (3) Weekly (4) Daily or almost daily 6. How often during the last year have you needed a first drink in the morning to get yourself going after a heavy drinking session? (0) Never (1) Less than monthly (2) Monthly (3) Weekly (4) Daily or almost daily 7. How often during the last year have you had a feeling of guilt or remorse after drinking? (0) Never (1) Less than monthly (2) Monthly (3) Weekly (4) Daily or almost daily 8. How often during the last year have you been unable to remember what happened the night before because you had been drinking? (0) Never (1) Less than monthly (2) Monthly (3) Weekly (4) Daily or almost daily 9. Have you or someone else been injured as a result of your drinking? (0) No (2) Yes, but not in the last year (4) Yes, during the last year 10. Has a relative or friend or doctor or other health worker been concerned about your drinking or suggested you cut down? (0) No (2) Yes but not in the last year (4) Yes, during the last year Record total of specific items here If total 8 or over, alcohol use disorder very likely. Scores above zero on items 4 through 6 indicate presence or emergence of alcohol dependence. Appendix 1: The AUDIT Questionnaire One standard drink is equal to… Half a pint of ordinary strength beer, lager or cider One small glass of wine One single measure of spirits One small glass of sherry One single measure of aperitifs Chapter 5.qxp 17/11/2006 11:01 Page 65 Review of the effectiveness of treatment for alcohol problems 66 1. MEN: How often do you have EIGHT or more drinks on one occasion? WOMEN: How often do you have SIX or more drinks on one occasion? (0) Never (1) Less than monthly (2) Monthly (3) Weekly (4) Daily or almost daily 2. How often during the last year have you been unable to remember what happened the night before because you had been drinking? (0) Never (1) Less than monthly (2) Monthly (3) Weekly (4) Daily or almost daily 3. How often during the last year have you failed to do what was normally expected of you because of drinking? (0) Never (1) Less than monthly (2) Monthly (3) Weekly (4) Daily or almost daily 4. In the last year has a relative or friend, or a doctor or other health worker been concerned about your drinking or suggested you cut down? (0) No (2) Yes, on one occasion (4) Yes, on more than one occasion Record total of specific items here A score >3 indicates probably hazardous drinking If the response to question one is “never”, the FAST test is negative. If the response is “weekly” or “daily or almost daily”, the FAST test is positive. Only ask questions two, three or four if the response to question one is “less than monthly” or “monthly”. If the response to questions two and three are ‘never’ and question four is ‘no’, the FAST test is negative. If there is any other response to questions two and three, then the FAST test is positive. Appendix 2: Fast Alcohol Screening Test (FAST) One standard drink is equal to… Half a pint of ordinary strength beer, lager or cider One small glass of wine One single measure of spirits One small glass of sherry One single measure of aperitifs Chapter 5.qxp 17/11/2006 11:01 Page 66 Screening for alcohol problems Circle number(s) for specific trigger(s); consider for all the top `10. 1. FALL (inc. trip) 2. COLLAPSE (inc. fits) 3. HEAD INJURY (inc. facial) 4. ASSAULT (inc. involved) 5. NON-SPECIFIC GI 6. “UNWELL” 7. PSYCHIATRIC (inc. overdose) 8. CARDIAC (inc. palpitations ) 9. SELF-NEGLECT 10. REPEAT attender Other (specifiy)_______________________________________________ After dealing with patient’s “agenda”, i.e. patient’s reason for attendance: 1. “We routinely ask all patients in A&E if they drink alcohol – do you drink?” If ‘yes’, go to question two. 2. “Quite a number of people have times when they drink more than usual; what is the most you will drink in any one day?” (Pub measures in brackets; home measures often x3!) Beer/lager/cider __ Pints (2) __ Cans (1.5) total units/day Strong beer/lager/cider __ Pints (5) __ Cans (4) _____________ Wine __ Glasses (1.5) __ Bottles (9) Fortified wine (sherry, Martini) __ Glasses (1) __ Bottles (12) Spirits (gin, whisky, vodka) __ Singles (1) __ Bottles (30) 3. If this is more than eight units/day for a man, or six units/day for a woman, does this happen: Everyday? = PAT +ve Dependent drinker Y/N (? Pabrinex) At least once a month? = PAT +ve Hazardous drinker Y/N 4. ‘Do you feel your current attendance in A&E is related to alcohol? Yes = PAT+ve No = PAT -ve If PAT +ve: “We gently advise you this drinking is harming your health. Would you like to see our health worker?” Yes/No – give leaflet Appendix 3: The Paddington Alcohol Test 67 Chapter 5.qxp 17/11/2006 11:01 Page 67 Review of the effectiveness of treatment for alcohol problems 70 6.2 Assessment tools 6.2.1 Context There are a number of assessment packages, which are discussed more fully in the section on routine follow-up (see section 6.3). They are useful for both initial assessment and follow-ups, as they can measure outcomes. It is not essential to use a standardised package but there is merit in having a core dataset that can be compared against population or other services’ data. In a guide for clinicians and researchers, the National Institute on Alcohol Abuse and Alcoholism (Allen and Wilson, 2003) describe the psychometric properties of 78 instruments, mainly North American, which may be used in the assessment, treatment and evaluation of people with alcohol problems (see also chapter five). The compilation is not exhaustive and includes scales that are not alcohol related. Waller and Rumball (2004) also describe a selection of instruments that are popular in the UK. Biochemical measures that are used for assessment may be added to the outcomes package (see chapter five). A good assessment tool should have both high reliability and validity: • Reliability refers to the extent to which measurements by the instrument can be reproduced, either from the same service user at different points in time (test-retest reliability), or from different raters who make the same measurements at the same point in time (inter-rater reliability). Reliability also refers to the internal consistency (the degree of inter-correlation) among the items making up a scale • Validity refers to the extent to which the instrument measures what it purports to be measuring. Validity comes in various forms: face, content, predictive, concurrent, discriminant and construct validity. Many instruments are available to assist the assessment and follow-up process. We describe the properties of some of the more popular instruments that tap into alcohol use and related problems. It is beyond the remit of this review to consider the many more instruments measuring constructs in domains other than substance use. 6.2.2 Evidence 6.2.2.1 Research diagnosis The Composite International Diagnostic Interview (CIDI) is a standardised and comprehensive interview schedule for the assessment of behavioural and psychological disorders, including “alcohol dependence” and “alcohol abuse”. It generates diagnoses according to the International Classification of Diseases (ICD-10; World Health Organization, 1993) and the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV; American Psychiatric Association, 1994). The CIDI must be administered or supervised by a qualified mental health professional who has received the recognised CIDI training. It is completed by both interview and self-report and takes approximately 70 minutes. A WHO international study (Ustun et al., 1999) compared the CIDI to two other comprehensive interview schedules. All three schedules had acceptable test-retest reliability and construct validity for alcohol dependence, but not for hazardous or harmful drinking. 6.2.2.2 Alcohol consumption Various methods can be used to record a service-user’s drinking behaviour (Sobell and Sobell, 1995). These include: • Quantity-frequency measures • Retrospective drinking diaries • Time-line follow-back method (TLFB) • Lifetime drinking history (LDH) • Self-monitored drinking logs. Quantity-frequency (Q-F) measures ask the service user to recall the “average” or “typical” frequency with which they consume alcohol, and the average or typical amount consumed per occasion. These are then multiplied to arrive at an overall level of consumption. Q-F measures can be supplemented by some measure of the variability of drinking. Retrospective drinking diaries record information about quantity, frequency and pattern of drinking by means of a detailed recall of actual drinking over a given time period, using prompts of time, place and drinking companions to elicit accurate recall. A given time period might be the previous week, or if that was atypical, the last typical week in the recent past. Chapter 6.qxp 17/11/2006 11:00 Page 70 Assessment and measuring treatment outcomes 71 The most thorough and sophisticated procedure is the time-line follow-back method (Sobell et al., 1988) which uses a calendar to elicit detailed information on drinking over an extended period of time. The lifetime drinking history (Skinner, 1982) is a formal method of obtaining information about the service user’s past drinking habits. It is likely that the passage of time will permit only key events to be noted, such as the introduction to drinking, periods of heavy use and of abstinence, and the onset of problems. Self-monitored drinking logs involve the recording of information by the service user on a daily or drink-by- drink basis. This method obviously relies less on memory than others and can be useful in monitoring progress in treatment, or identifying high-risk relapse situations. An assessment of alcohol consumption will also be a convenient time to enquire about polydrug use, since some service users will use and may have problems with other drugs. A profile of use of all common psychoactive substances (including tobacco) can be obtained. Research reveals many problems with Q-F measures, which have been reviewed by Sobell and Sobell (1995). The main problems are that they tend to underestimate consumption and miss episodes of binge drinking. However, Q-F measures can provide an easily administered and quick assessment of drinking, if time is limited. Retrospective diary methods are generally more accurate than Q-F measures, particularly with respect to binge drinking (Redman et al., 1987; Shakeshaft, Bowman and Sanson-Fisher, 1998). They are less accurate in estimating low levels of consumption, but this is clearly not a serious limitation in treatment assessments. There is an extensive body of research to support the reliability of the TLFB with a variety of types of drinker (Sobell and Sobell, 1995). Completion of the TLFB calendar with the service user can also provide useful clinical information. It may be too time-consuming for routine clinical purposes but an understanding of the principles of the method may be useful for interviewers. The LDH shows reasonably high reliability as an aggregate index of drinking over a lifespan but lacks accuracy for more recent periods of time (e.g. the year before interview) (Skinner and Sheu, 1982) and other methods should be used here. No research seems to have been conducted on the reliability and validity of self- monitoring logs. Possible problems with compliance are obvious (Sobell and Sobell, 1995). 6.2.2.3 Alcohol dependence This is an important assessment domain and one where the advantages of standardised measurement are probably greatest. Edwards, Marshall and Cook (2003) state that while the degree of dependence is not all there is to assessment, its measurement is of great practical importance, and is essential to treatment planning. A number of well-known instruments exist, with various advantages and disadvantages (Davidson, 1987), and at least one of these should be used. Severity of Alcohol Dependence Questionnaire The Severity of Alcohol Dependence Questionnaire (SADQ) was one of the first measures of alcohol dependence to be developed (Stockwell et al., 1979) and is based on the elements of the alcohol dependence syndrome described by Edwards and Gross (1976). It consists of 20 items and is divided into five sections referring to: • Physical withdrawal symptoms • Affective withdrawal symptoms • Craving and relief drinking • Typical daily consumption • Reinstatement of dependence after a period of abstinence. The SADQ is widely used in the UK and is often employed to give advice to service users on the suitability of abstinence or moderation goals. A score of 30 or above on the SADQ is conventionally taken to indicate severe dependence (Stockwell, Murphy and Hodgson, 1983). The SADQ takes about five minutes to complete. Stockwell et al. (1994) developed a 16-item version suitable for use in community samples – the SADQ-C. This comes with the short Impaired Control Questionnaire to measure the extent to which respondents believe their drinking is out of control. Rather than asking about a “recent period” of heavy drinking, as in the original SADQ, the SADQ-C asks about drinking in the past three months. The complete SADQ-C takes between five and ten minutes to complete and, as with all the measures discussed in this section, does not require any special training to administer it. In its original validation study (Stockwell et al., 1979), the SADQ showed significant inter-correlations between the five sections of the questionnaire. There was also high concordance between score on the SADQ and a Chapter 6.qxp 17/11/2006 11:00 Page 71 Review of the effectiveness of treatment for alcohol problems 72 clinician’s rating of degree of dependence. The SADQ also showed good test-retest validity and significant correlations with observer ratings of withdrawal severity and narrowing of the drinking repertoire (Stockwell, Murphy and Hodgson, 1983). The good reliability and validity of the SADQ was independently confirmed among a sample of Irish alcohol misusers (Meehan, Webb and Unwin, 1985). It has been suggested that a lower cut- point for the designation of severe dependence may be appropriate for women (Dawe et al., 2002a). In its development study (Stockwell et al., 1994), the SADQ-C showed good reliability and validity in an Australian general population sample. Alcohol Dependence Scale The Alcohol Dependence Scale (ADS) is a 25-item self- report questionnaire based again on the alcohol dependence syndrome (Skinner and Allen, 1982). It was derived from analysis of the larger Alcohol Use Inventory and measures loss of behavioural control, psycho- perceptual withdrawal symptoms, psychophysical withdrawal symptoms and obsessive-compulsive drinking style. Several studies among a diversity of alcohol misusers have shown the ADS to have good reliability and validity (Dawe et al., 2002a). The ADS is more used in North America than in the UK Leeds Dependence Questionnaire The Leeds Dependence Questionnaire (LDQ) is a more recent, ten-item instrument that offers a generic measurement of dependence on any psychoactive substance (Raistrick et al., 1994). If preferred, the wording can be made specific to alcohol or any other drug. The LDQ is based on a psychological understanding of dependence and so does not directly measure symptoms of tolerance and withdrawal that are a function of recent drinking. It was designed to be sensitive to change over time and to be sensitive through the range of mild to severe dependence. As suggested by its length, the LDQ can be completed more quickly than the other instruments discussed here. The development study of the LDQ (Raistrick et al., 1994) showed that the instrument measured a single construct, showed high test-retest reliability and had satisfactory concurrent, discriminant and construct validities. Subsequently, Heather et al. (2001) examined the psychometric properties of the LDQ among a large sample of service users attending treatment for substance use disorders. The satisfactory reliability and validity of the instrument were confirmed and it was shown to give a robust and sound assessment of dependence across a range of substances. It has been shown to measure dependence during periods of abstinence (Tober, 2000). Ford (2003) has demonstrated the clinical usefulness and validity of the LDQ in a population of problem drinkers who also have a mental illness. The cut-points are <10 for low dependence, 10- 22 for medium dependence and >22 for high dependence. 6.2.2.4 Alcohol-related problems The degree to which service users experience alcohol- related problems is a different matter from their degrees of alcohol dependence. Dependence and problems are correlated but are conceptually independent areas of functioning (Edwards et al., 1977). It is possible for someone to have a severe level of dependence but only few and mild problems, and vice-versa. Many earlier instruments purporting to measure problems in fact confused alcohol-related problems and dependence. In a thorough assessment of alcohol-related problems, the whole range of negative consequences that might have been experienced by the service user should be covered, including medical, psychological, financial, legal, vocational, social, marital and other interpersonal problems. Alcohol Problems Questionnaire The Alcohol Problems Questionnaire (APQ) (Drummond, 1990) is a “pure” measure of alcohol-related problems, developed in the UK. It covers eight problem areas: friends, money, police, physical, affective, marital, children and work. The last two of these apply only to service users with children or in work. Subscale scores are calculated for each area and a common score based on 23 items is derived. Williams and Drummond (1994) reported a high test-retest reliability for the APQ common score and moderate-to-high reliabilities for subscale scores. There was a moderate but highly significant correlation with the SADQ and levels of dependence were the strongest predictors of APQ score compared to a range of other variables. 6.2.2.5 Motivation to change An understanding of the service user’s motivation to change drinking behaviour is a key to effective treatment Chapter 6.qxp 17/11/2006 11:00 Page 72 Assessment and measuring treatment outcomes 75 Many other instruments exist for use in conjunction with cognitive behavioural therapy and these may be found in Allen and Wilson (2003). 6.2.3 Conclusions • There are many instruments with good psychometric properties that can be combined to construct an assessment package; packages should also be suitable for outcome ratings (see chapter 15) (I) • The CIDI provides a thorough but time-consuming assessment with satisfactory reliability and validity for diagnosing alcohol dependence according to ICD-10 or DSM-IV criteria (II) • Q-F measures of alcohol consumption can be used when time is limited but they are likely to be inaccurate to varying degrees (III) • Retrospective drinking diaries offer the most reliable method of recording alcohol consumption in routine clinical practice, particularly using time-line follow-back (II) • Several reliable and valid instruments exist for the measurement of alcohol dependence and one of these should be used in assessment (II) • The APQ is the instrument of choice for the measurement of alcohol-related problems in the UK (II) • The RCQ and RCQ (TV) provide brief methods of assessing a service user’s stage of readiness to change drinking behaviour with moderately good psychometric properties (II) • A collection of instruments are available for use in conjunction with cognitive behavioural therapy (II) 6.3 Routine follow-up 6.3.1 Context Actual clinical outcomes are the summation of a number of influences that include how well treatment has been delivered, how good the specific treatment was, the quality of clinical governance controls in the agency and the quality of organisational support. Outcomes can be an important component of performance management but their proper use requires resources and methodological knowledge and skills (Tonigan, 2003, pp 219–233). To be meaningful, outcome measures need to be accompanied by a description of the cohort of service users in question, a follow-up of a representative sample from the original cohort and a chart recording the nature of interventions given and the reasons for drop-outs. The change measures – for example, dependence scores or percentage of days abstinent – can be compared in a number of ways: • Simple statistical terms, such as comparing mean scores at baseline and follow-up • Clinically significant change (Jacobson et al., 1999; Tober, 2000, pp 182-191) • Categorical terms, for example using ICD-10. 6.3.2 Evidence Depending upon definition, as many as 70 per cent of service users new into treatment will have relapsed at six month follow-up. This does not imply that these individuals are lost to treatment; indeed the lapse or relapse can often be used to therapeutic advantage. Changes in drinking behaviour tend to occur in the first three months of treatment and the benefits across a range of outcome domains are typically maintained through to 12 months (Babor et al., 2003b; Weisner et al., 2003). It follows that a three-month follow-up will give the best indication of treatment effectiveness, while a 12 month follow-up will give a better idea of the overall benefits of treatment, albeit shaped by an individual’s characteristics and circumstances. At 12 months it is possible that less than 30 per cent of new service users will still be in contact with an agency; however, it is possible to boost this to at least 80 per cent by using trained follow-up staff (Cottler et al., 1996; Tober et al., 2000). 6.3.3 Conclusions • Routine evaluation of treatment outcomes is feasible but requires follow-up staff and access to statistical advice (II) • Reporting clinically significant change is a strict test of outcome, which gives a good indication of improvement meaningful at an individual level (II) • There is a logic to undertaking follow-ups three months and 12 months after entering treatment and then again annually (IV). Chapter 6.qxp 17/11/2006 11:00 Page 75 Review of the effectiveness of treatment for alcohol problems 76 6.4 Assessment packages 6.4.1 Context Selecting suitable outcome measures for an agency is not as simple as might be supposed. There are a number of reasons: 1 Agencies need to have ownership of their data collection, otherwise motivation will be lacking and data quality poor 2 Different tiers of services will have different assessment requirements, different commitments to aftercare and follow-up 3 Within tiers, agencies may have different treatment objectives 4 Many outcome measurement tools lack adequate validation. That said, alcohol consumption is common to all services and outcomes (Sobell and Sobell, 2003, pp75-99) and is a logical starting point for a small common dataset. The selection of additional outcome measures might take account of whether the measures: • Are universal – not constrained by any particular substance or social group • Have proven validity and reliability and have published psychometric properties • Are sensitive to change • Have easy readability and neutral language • Are practitioner-completed (subject to bias), self- completed (free of practitioner bias), or a mixture of both. The alcohol research community has, to some extent, already settled the debate by choosing common measures to compare results across major trials (e.g., UKATT Research Team, 2001; Babor and Del Boca, 2003). Data from these trials provide useful comparison groups for clinical services that choose the same outcome measures. 6.4.2 Evidence Here we describe four treatment outcome packages and one scale, all of which cover the key outcome domains and are widely used for both research and clinical purposes. These packages have been designed and tested with a particular function in mind but there is no barrier, in principle, to designing a local package made up from a selection of the individual scales available. 6.4.2.1 Comprehensive Drinker Profile The Comprehensive Drinker Profile (CDP) (Miller and Marlatt, 1987) was designed to provide clinically useful information on the level of alcohol consumption, drinking patterns, alcohol-related problems and socio- demographic background. It is accompanied by: • A Brief Drinker Profile, for use when time is limited, or when the client is reluctant to complete the full profile • A Follow-Up Drinker Profile, for recording outcome after treatment in a comparable form to pre-treatment measurements • A Collateral Interview Form, for structuring corroborative interviews with family or friends. A trained interviewer must administer the CDP. It contains 88 items and takes two hours to complete. No independent studies of the CDP appear to have been conducted. Dawe et al. (2002a) state that the CDP family of instruments “… make a good assessment tool for clinicians working with clients who present with alcohol problems” (p.66). Sobell and Sobell (1995) write that the CDP “… provides a consistent baseline dataset for treatment planning with structured parallel interviews that can be used for follow-up or with collaterals” (p.68). 6.4.2.2 The Addiction Severity Index The Addiction Severity Index (ASI) (McLellan et al., 1980) is a multidimensional structured interview for assessing dependence and problems across the full range of substance use disorders. It consists of 200 items and seven subscales measuring alcohol consumption, other drug use, medical problems, psychiatric status, family- social problems, employment problems and legal difficulties. The ASI has been updated over the years and is now in its fifth revision (McLellan et al., 1992). It is widely used in treatment and research and has been translated into all major languages. Scoring takes account of subjective ratings of severity by service users and objective evidence to arrive at overall severity ratings. The ASI need not be given by a mental health professional but needs training in its administration. The time required is estimated at 50–60 minutes. The time Chapter 6.qxp 17/11/2006 11:00 Page 76 Assessment and measuring treatment outcomes 77 frame covers the past 30 days of recent use, otherwise it refers to lifetime use. The ASI has been widely used in both clinical and research settings and there has been supporting validation (see Rosen et al., 2000; Leonhard et al., 2000). Several studies have examined the psychometric properties of the ASI and it has been generally found to have good reliability and validity as an indicator of treatment outcome (Dawe et al., 2002a). However, Makelä (2004), who reviewed 37 studies of its psychometric performance, has recently questioned the reliability and validity of the ASI. He concluded: • Inter-rater and test-retest reliabilities of severity ratings and composite scores varied from excellent to unsatisfactory • High internal consistencies were reported regularly for only three of the seven composite scores • The remaining four composite scores (employment status, drug use, legal status and family-social relations) have low consistencies in at least four different studies • Indices of construct validity are often low. 6.4.2.3 The Maudsley Addiction Profile The Maudsley Addiction Profile (MAP) (Marsden et al., 1998) is a brief structured interview for treatment outcome research, with 28 items covering substance use, health risks, health symptoms (ten-item scale), psychological symptoms (ten-item scale), social functioning, and criminal activity. Completion time is approximately 20 minutes. Most of the measures in MAP are standardised on a 30-day time frame The MAP is used for both research and clinical purposes in the UK. The instrument can be added to, for example, with a measure of substance dependence. The MAP has been field tested in a European context in combination with the Treatment Perceptions Questionnaire, a standalone satisfaction rating (Marsden et al., 2000a), and this extends the completion time slightly (Marsden et al., 2000b). 6.4.2.4 RESULT RESULT (Raistrick and Tober, 2003) combines the substance misuse and physical health items of MAP with self-completion measures of dependence (Raistrick et al., 1994; Ford, 2003), psychological morbidity (Evans et al., 2002), and social satisfaction (based on Corney and Clare, 1985). An alternative to social satisfaction would be the Alcohol Problems Questionnaire (Drummond, 1990), which is commonly used in trials but for clinical purposes falls down on the universality test (see section 6.4.1). The package can be computerised and was designed for routine use in clinical services, combining alcohol and other drugs. Completion time is approximately ten minutes for the substance misuse history and ten minutes for the self-completion questionnaires. The time frame is 30 days. RESULT is used for both research and clinical purposes in the UK. 6.4.2.5 The Christo Inventory for Substance Misuse Services The Christo Inventory for Substance Misuse Services (CISS) (Christo et al., 2000) is a single-page outcome evaluation tool completed by the service user’s therapist from direct interviews, or retrospectively from case notes. It is a ten-item scale with each item scored zero to two. The items cover social functioning, health, risk behaviour, psychological wellbeing, occupation, criminal activity, substance use, support, treatment compliance and therapeutic alliance. Completion time is approximately ten minutes. The time frame is the last 30 days. The CISS has high face validity and is used in clinical services across the UK. 6.4.3 Conclusions • The reliability and validity of assessment packages have not been independently examined (other than one meta-analysis on the ASI) and so the evidence to support standard assessment packages is weak (IV) • The CDP family of instruments provide a lengthy but clinically useful and thorough assessment of alcohol problems. The reliability and validity have not been independently examined (IV) • The ASI is a widely used, comprehensive assessment tool but reliability and validity have come into question. MAP or RESULT are alternatives but have not been independently examined (IV) • Measures that will be useful for routine clinical use can often be taken from major clinical trials (IV) • There is ample scope to mix different scales for agencies to create a preferred package drawing on commonly used assessment tools (see chapter five) (IV). Chapter 6.qxp 17/11/2006 11:00 Page 77 Review of the effectiveness of treatment for alcohol problems 80 making neat classifications among the wide variety of treatments and interventions found in the literature on alcohol problems. 7.2.2 Evidence Together with studies categorised as motivational enhancement in the Mesa Grande, there is a very large body of research evidence on alcohol brief interventions, including at least 56 controlled trials of effectiveness (Moyer et al., 2002). There have been at least 14 meta- analyses or systematic reviews, using somewhat different aims and methods, of research on effectiveness of brief interventions (Bien, Miller and Tonigan, 1993; Freemantle et al., 1993; Kahan, Wilson and Becker, 1995; Wilk, Jensen and Havighurst, 1997; Poikolainen, 1999; Irvin, Wyer and Gerson, 2000; Moyer et al., 2002; D’Onofrio and Degutis, 2002; Berglund, Thelander and Jonsson, 2003; Emmen et al., 2004; Ballesteros et al., 2004a; Whitlock et al. 2004; Cuijpers, Riper and Lemmens, 2004; Bertholet et al., 2005). All these have reached conclusions, in one form or another, favouring the effectiveness of brief interventions in reducing alcohol consumption to low-risk levels among hazardous and harmful drinkers. In the most comprehensive and well-designed meta- analysis in this area (Moyer et al., 2002), the studies were divided into 34 opportunistic interventions carried out in generalist settings among individuals not seeking treatment for alcohol problems and 20 specialist brief interventions among those who were seeking treatment. From the first group of studies, which are of interest in this chapter, small to medium aggregate effect sizes in favour of brief interventions emerged across different follow-up points. At follow-ups of between three and six months inclusive, the effect for brief interventions compared to control conditions was significantly larger when alcohol misusers showing more severe alcohol problems were excluded from the analysis. In addition, the majority of studies of brief interventions have excluded individuals showing significant levels of dependence, so that the findings apply mainly to service users with no or only mild dependence. Therefore, service users with moderate or severe levels of dependence should routinely be referred for specialist treatment; it is possible that a few of these service users may benefit from a brief intervention but research suggests that they should at least be offered referral to and encouraged to attend specialist services for treatment of alcohol dependence. Other evidence-based reviews consulted for this document found brief interventions to be effective: • The Swedish Technology Assessment review (Berglund, Thelander and Jonsson, 2003) concluded: “In most of the studies [of brief intervention for secondary prevention] a significant effect of brief intervention has been shown in follow-ups for up to two years. The treatment effect is of the same magnitude as that achieved with many common medical treatments for chronic conditions” (p38) • The Australian systematic review (Shand et al., 2003a) concluded that “opportunistic brief interventions are effective in reducing alcohol consumption in problem drinkers with low levels of dependence” (p44) • The Scottish review (Slattery et al., 2003) was concerned exclusively with service users being treated in specialist services following alcohol detoxification. It concluded that brief interventions were not recommended for use in this population, as research had failed to shown any benefit. There is mixed evidence on longer-term effects of brief interventions: • A trial based in family medicine in Wisconsin, USA reported continuing benefits for alcohol use, binge drinking episodes and frequency of excessive drinking among recipients of brief interventions compared with controls four years after intervention (Fleming et al., 2002) • An Australian study reported that the benefits of receiving brief interventions had disappeared after ten years (Wutzke et al., 2002) and it was suggested that booster sessions would be necessary to maintain the effect over this period of time • A 10-16 year follow-up sample recruited in a pioneering Swedish study carried out as part of a health screening programme showed reduced mortality in the intervention group (Kristenson et al., 2002) but it is questionable whether this study can be regarded as relevant to brief intervention because of the length and duration of the original intervention sessions. More research is clearly needed, particularly in the UK, on the longer-term effects of brief interventions. Chapter 7.qxp 17/11/2006 11:00 Page 80 Brief interventions 81 There is some evidence that brief interventions reduce alcohol-related mortality (Cuijpers, Riper and Lemmens, 2004), albeit from a small number of studies. Moyer et al. (2002) also reported that brief interventions were effective on a composite of various drinking-related outcomes, including measures of alcohol-related problems. There is also direct evidence from an Australian study in general practice that brief interventions are effective in reducing alcohol-related problems among those who receive them (Richmond et al., 1995). More studies of the effects of brief interventions other than on alcohol consumption itself, including effects on mortality, general adjustment and alcohol problems, would be useful. The issue of the cost-effectiveness of brief interventions will be addressed in chapter 14. 7.2.3 Conclusions • Brief interventions, of various forms and delivered in a variety of settings, are effective in reducing alcohol consumption among hazardous and harmful drinkers to low-risk levels (IA) • Effects of brief interventions persist for periods up to two years after intervention and perhaps as long as four years (IB) • Booster sessions may be necessary to maintain the effect for longer periods of time, although more research is needed on the longevity of the effects of brief interventions (IB) • Brief interventions are effective in reducing alcohol- related problems among harmful drinkers (IIA), although more research would be useful • There is some evidence that they are effective in reducing alcohol-related mortality, although more research is needed (IA) • There is no evidence that opportunistic brief interventions are effective among people with more severe alcohol problems and levels of dependence, i.e. among moderately and severely dependent drinkers (IA) and such service users should be encouraged to attend specialist treatment services. 7.3 Brief interventions in primary healthcare 7.3.1 Context There are many advantages in delivering brief interventions in primary healthcare, due mainly to the access it provides to the majority of the general population, the absence of stigma attached to attending primary care facilities, the presence of “teachable moments” in consultations about alcohol-related illnesses, and the generally high credibility in the community of GPs and other primary care professionals (Babor, Ritson and Hodgson, 1986). 7.3.2 Evidence Studies by Wallace, Cutler and Haines (1988) and by Anderson and Scott (1992) in the UK established the effectiveness of brief interventions delivered by general practitioners in reducing the proportion of patients drinking above medically recommended guidelines. The public health potential of GP-based brief interventions was highlighted by Wallace et al. when they estimated, on the basis of their findings, that routine and consistent implementation of their intervention program by general practitioners throughout the United Kingdom would result in a reduction from hazardous or harmful to low-risk levels of the drinking of 250,000 men and 67,500 women each year. There have been five systematic reviews with meta- analysis specifically focused on the effectiveness of brief interventions in primary healthcare (Kahan, Wilson and Becker, 1995; Poikolainen, 1999; Ballesteros et al., 2004a; Whitlock et al., 2004; Bertholet et al., 2005). The most recent of these (Bertholet et al., 2005) concluded that brief interventions are effective in reducing consumption among both men and women at six and 12 months following intervention. This review was confined to studies carried out in more naturalistic conditions of primary healthcare, excluding those studies that used patient lists, registers or specially arranged screening sessions, and is therefore more relevant to real world conditions of general practice than other reviews. Another recent review (Ballesteros et al., 2004a) concluded that their meta-analysis, although indicating a smaller effect size than reported in previous reviews, Chapter 7.qxp 17/11/2006 11:00 Page 81 Review of the effectiveness of treatment for alcohol problems 82 nevertheless supported the moderate effectiveness of opportunistic brief interventions. Yet another recent review, by the US Preventive Task Force (Whitlock et al., 2004), found that “… brief counselling interventions for risky/harmful alcohol use among adult primary care patients could provide an effective component of a public health approach to reducing risky-harmful alcohol use.” (p557). With regard to gender, Ballesteros et al. (2004b) found in their meta-analysis that, despite indications from previous research that brief interventions may be less effective among women than men (e.g. Scott and Anderson, 1991; Anderson and Scott, 1992), there was no evidence of any difference in response between genders. Fleming et al. (1999) reported that brief interventions delivered in general practice were effective too among adults over 65 years old. The effect size of brief interventions is more understandable in terms of number needed to treat (NNT – the number of hazardous of harmful drinkers that need to receive intervention for one to reduce drinking to low- risk levels). The latest estimate of NNT for brief interventions is about eight (Moyer et al., 2002). This compares favourably with NNT for advice to quit smoking which has an NNT of 20, although this improves to about ten with the addition of nicotine replacement therapy (Silagy and Stead, 2003). In a sense, NNT underestimates the full effectiveness of brief intervention since, even if the drinker does not immediately reduce drinking, it may plant a seed that later becomes an active effort to cut down, or – in other words – the beginning of a movement along the cycle of change. In any event, as with smoking cessation advice, the NNT for alcohol brief interventions indicates that, if routinely implemented in primary healthcare, its potential to reduce alcohol-related harm in the population is very large. 7.3.3 Conclusions • Opportunistic brief interventions delivered to hazardous and harmful drinkers in primary healthcare are effective in reducing alcohol consumption to low risk levels (IA) • The public health impact of widespread implementation of brief interventions in primary healthcare is potentially very large (IB) • NNT for alcohol brief interventions in primary healthcare is about eight and this compares favourably with advice to quit smoking (IA) • Brief interventions in primary healthcare are equally effective among men and women (IA) • Brief interventions in primary healthcare are effective among older adults (IB). 7.4 Brief interventions in the general hospital 7.4.1 Context In some ways, the general hospital ward offers a setting more conducive to brief interventions than primary healthcare, mainly because patients have more time available for screening and intervention. There is abundant evidence that many types of hospital ward contain high numbers of hazardous and harmful drinkers, especially among males, not to mention alcohol dependent patients. Depending on the definitions used, it is estimated up to 40 per cent of male patients are alcohol misusers (Royal College of Physicians, 1987). There has been one meta-analysis of opportunistic brief interventions in the general hospital setting (Emmen et al., 2004). This was based on eight studies, most of which the authors regarded as having methodological weaknesses. Only one study, with a relatively intensive intervention and a short follow-up period, showed a significantly large reduction in alcohol consumption in intervention groups (Maheswaran et al., 1992); this was conducted among hypertensive patients. The conclusion of the Emmen et al. review was that: “Evidence for the effectiveness of opportunistic brief interventions in a general hospital setting for problem drinkers is still inconclusive.” (p322). There are reasons to believe that this conclusion may be unduly pessimistic: • An early study in Edinburgh (Chick, Lloyd and Crombie,1985) reported that a one-hour intervention on the ward by a nurse was effective in reducing alcohol-related harm in the one-year follow-up period. This harm-reduction effect of brief intervention, in the absence of significant changes to alcohol consumption, has been reported in other studies of Chapter 7.qxp 17/11/2006 11:00 Page 82 Brief interventions 85 7.6.1.2 Needle exchange programmes Stein et al. (2002) in the USA investigated the effects of a brief motivational intervention for reducing alcohol use among service users of a needle exchange programme. Participants randomised to the intervention received a one-hour session of motivational interviewing with a booster session one month later, while controls received usual care. At six-month follow-up, participants in the intervention group showed significantly greater reductions in consumption, but the authors state that the optimal length of intervention in this setting deserves further study. 7.6.1.3 Prenatal care In the USA, Chang et al. (1999b) assessed the impact of brief interventions on ante partum alcohol consumption among pregnant women receiving prenatal care. Both intervention and assessment–only control participants had reduced consumption at follow-up, but there were no significant differences between groups. Considering the importance of reducing excessive alcohol consumption among pregnant women, more studies of intervention in this context are warranted. 7.6.1.4 Somatic outpatient clinics In a small study in Norway, Persson and Magnusson (1989) examined the effectiveness of a brief and early intervention among patients at a “somatic outpatient clinic” who had not yet experienced medical or social negative consequences from their alcohol misuse. At follow-up interviews over 12 months, participants in the intervention group showed decreased consumption, liver enzyme readings and sickness days compared with controls. The authors conclude that their early intervention programme was effective, carried out at low cost and received a positive response from patients. 7.6.1.5 General population health screening programmes Pioneering studies of brief interventions in Scandinavian countries (Kristenson et al., 1983; Nilssen, 1991) were carried out as part of general population health screening programmes. In general terms, these studies provide good evidence for the effectiveness of these interventions, although, as we have noted, it is doubtful whether they can be considered brief. 7.6.2 Conclusions • There is some evidence that brief interventions are effective in producing short-term reductions in alcohol consumption among psychiatric patients with mid- range psychiatric disorders (IB) • There is some evidence that brief interventions are effective in reducing the alcohol consumption of heavy drinking service users in needle exchange programmes (IB) • There is no evidence as yet that brief interventions reduce alcohol consumption among pregnant women (IB) • There is some evidence that brief interventions are effective among patients attending outpatient clinics for somatic disorders (IB) • Scandinavian trials of intervention delivered as part of general population health screening programmes showed positive effects, though these interventions were more intensive than those normally considered “brief” (IB). 7.7 Brief interventions in educational establishments 7.7.1 Evidence A series of studies by G Alan Marlatt and colleagues from the University of Washington tested the effectiveness of brief interventions on campus among heavy drinking college students (Baer et al., 1992; Marlatt et al., 1998; Baer et al., 2001). Earlier studies used a condensed form of cognitive-behavioural therapy but more recent work has focused on brief motivational interviewing. In the most recent study (Baer et al., 2001), heavy drinking students in their freshman year were randomly allocated to an intervention group that received individual motivational feedback based on a prior assessment, followed by mailed feedback derived from six-month and one-year follow-up contacts. At a two-year research follow-up, the intervention group showed greater reductions in drinking and harmful consequences compared to a non-intervention control group. The intervention group continued to report more alcohol problems that a matched, natural history comparison group not showing heavy drinking. However, the decline in problems over time suggested that the effects of brief Chapter 7.qxp 17/11/2006 11:00 Page 85 Review of the effectiveness of treatment for alcohol problems 86 motivational intervention were added to maturational processes. At a later four-year follow-up, these trends were confirmed and the authors concluded that brief interventions for high-risk college drinkers “can achieve long-term benefits even in the context of maturational trends” (p1310). Borsari and Carey (2000) randomised college student binge drinkers to a one-session motivational intervention or a no-treatment control group. The intervention provided students with feedback regarding their personal consumption, perceived drinking norms, alcohol-related problems, situations associated with heavy drinking and alcohol expectancies. At six-week follow-up, the brief intervention group showed significant reductions in number of drinks per week and frequency of binge drinking in the past month. In a recent study carried out in ten further education colleges in inner London, McCambridge and Strang (2004) evaluated the effects of a single one-hour, individual session of motivational interviewing on students’ (16-20 years) drug use, including alcohol, cigarettes and cannabis. Control group participants received education as usual. At a three-month follow-up, students who had received interventions showed significantly greater reductions in alcohol and cannabis use, an effect that was greater among heavier users of both drugs. This effect had almost entirely disappeared at a later 12-month follow-up (McCambridge and Strang, 2005), although the authors suggest that this was mainly due to an improvement in the control group, not a return to baseline levels in the intervention group (see also Miller, 2005). 7.7.2 Conclusion • Brief motivational interventions are effective in reducing levels of alcohol consumption and frequency of binge drinking among heavy-drinking college students (IB). 7.8 Brief interventions in other non-medical settings 7.8.1 Evidence 7.8.1.1 Social work Given the extensive contribution of excessive drinking to the social work caseload, social services would seem to provide an important opportunity for brief interventions. However, although there has been plenty of advice on how social workers should respond to alcohol problems in their service users (e.g., Alaszewski and Harrison, 1992), there have been no controlled evaluations of brief interventions in a social work context. 7.8.1.2 Criminal justice system It would be possible to implement brief interventions in prisons, probation settings and even police stations, as well as establishing special types of intervention for specific groups such as drink-driving offenders. There appear to have been no attempts as yet to evaluate the effectiveness of such possibilities in the UK. However, the Government intends to fund pilot research into the practical implementation of brief interventions in criminal justice settings. 7.8.2 Workplace There has been some development and evaluation of workplace brief interventions in Australia (Richmond et al., 1992) and the US (Higgins-Biddle and Babor, 1996), but no attention to this possibility in the UK. 7.8.3 Conclusions • Studies are needed of the effectiveness of brief interventions in social work settings (IV) • Studies are needed of the effectiveness of brief interventions in various settings within the criminal justice system (IV) • UK research is needed on the effectiveness of brief intervention in the workplace (IV). 7.9 Simple brief interventions 7.9.1 Context So far in this review, we have spoken of brief intervention as an umbrella term. It is now time to distinguish between simple and extended brief intervention. One of the most influential studies in this area was the WHO clinical trial in primary healthcare (Babor and Grant, 1992). The basic five minutes of advice found to be effective in this trial can be used by busy physicians or other healthcare workers who would not have time for a more prolonged intervention. The 20 minutes of assessment that preceded the WHO intervention can be replaced by the Chapter 7.qxp 17/11/2006 11:00 Page 86 Brief interventions 87 results of screening tests and the clinician’s knowledge of the person. In addition to research evidence, there are also logistical reasons to support the implementation of simple brief interventions for hazardous and harmful drinkers across the health system. Given the huge numbers of hazardous and harmful drinkers in the general population, it is inconceivable that all could be offered any more prolonged intervention that a few minutes of simple advice. Even if they are in the pre-contemplation stage of change and do not wish help to cut down or quit drinking, hazardous and harmful drinkers have a right to receive information that their drinking places them at risk of developing medical and social problems and on the limits for sensible drinking. Besides this basic information, simple brief interventions should include the following, all of which have support from the research literature and derive from the FRAMES acronym originally described by Miller and Sanchez (1994): • Structured and personalised feedback on risk and harm • Emphasis on the patient’s personal responsibility for change • Clear advice to the patient to make a change in drinking • A menu of alternative strategies for making a change in behaviour • Delivered in an empathic and non-judgmental fashion • An attempt to increase the patient’s confidence in being able to change behaviour (self-efficacy). Simple brief interventions should also include goal-setting (e.g. start date and daily or weekly limits for drinking), written self-help material for the patient to take away – containing more detailed information on consequences of excessive drinking and tips on cutting down – and arrangements for follow-up monitoring. Competence in delivering simple brief interventions does not need extensive training and one or two sessions of instructive and practical training should suffice. Assuming the necessary levels of interpersonal skills are present, training should cover the rationale and aims of brief interventions, the types of drinkers to whom they should be offered, the benefits for health and welfare that are likely to follow, an introduction to the stages of change model and perhaps some role-play practice in delivering advice with feedback on performance. 7.9.2 Evidence The WHO trial was an international collaboration involving ten countries and 1,655 heavy drinkers recruited from a combination of various, mostly medical settings (Babor and Grant, 1992). This clearly established that, among males, an intervention consisting of five minutes simple advice based on 20 minutes of structured assessment is effective in reducing alcohol consumption, with concomitant improvements in health. Among women, participants receiving simple advice and those just receiving an assessment both reduced consumption and there was no significant difference between these groups. However, later research and analysis have shown that women may be more responsive to brief intervention than men (Fleming et al., 1997), suggesting that women in the WHO trial showed a positive response to receiving an alcohol-related assessment only. Simple, structured advice should ideally be offered to all hazardous and harmful drinkers who screen positive for or are otherwise identified as such. As first suggested by Wallace, Cutler and Haines (1988), in addition to benefit for individuals, the public health impact of a widespread implementation of simple brief intervention is likely to be very large. 7.9.3 Conclusion • Simple brief interventions consisting of simple, structured advice are effective in reducing alcohol consumption and improving health status among hazardous and harmful drinkers encountered in healthcare settings (IB). 7.10 Extended brief interventions 7.10.1 Context An extended brief intervention typically takes 20-30 minutes to deliver and can involve a small number of repeat sessions. It should be directed towards harmful drinkers whose levels of alcohol-related harm indicate a need for it and who are willing to accept it. It may also be suitable for hazardous drinkers in the contemplation stage of change, who are ambivalent about their drinking and Chapter 7.qxp 17/11/2006 11:00 Page 87 Review of the effectiveness of treatment for alcohol problems 90 Some of these barriers could be fairly easily overcome. Screening and intervention materials are available and need only to be widely disseminated; appropriate training could be provided; evidence that brief interventions are effective could be better communicated to health professionals. Some of the negative attitudes to this work could be changed by emphasising the difference between the targets for brief intervention and the management of severely dependent individuals with serious problems, and by facilitating arrangements for referring the latter group to specialist treatment. Fear of offending patients could be partly reduced by evidence that most patients expect GPs and nurses to enquire about their drinking in appropriate circumstances and see this as a legitimate part of medical practice (Wallace and Haines, 1984; Richmond et al., 1996; Rush, Urbanoski and Allen, 2003; Hutchings et al., 2006). Probably the most difficult obstacles are those to do with lack of time and of reimbursement for this work. Research by Kaner and colleagues, as part of Phase III of the WHO Collaborative Project on Brief Interventions for Hazardous and Harmful Alcohol Use, has shown that telemarketing is the most cost-effective means of disseminating brief intervention programmes in primary healthcare (Lock et al. 1999). The same research team randomised GPs to one of three groups: (i) training and support; (ii) training and no support; (iii) a control group receiving no training or support (Kaner et al., 1999b). Results showed that trained and supported GPs implemented a screening and brief intervention programme more extensively and systematically than those who received training alone or the control group and that this was a cost-effective strategy for encouraging GPs to use the programme on a longer-term basis. This was confirmed in a subsequent analysis by Anderson et al. (2003; 2004a) of data from several countries taking part in this WHO collaborative study. This showed that, when GPs and nurses are adequately trained and supported, screening and intervention activity increases. GPs who expressed more confidence in working with alcohol problems and who reported greater therapeutic commitment to this work were more likely to manage patients with alcohol-related harm (Anderson et al., 2003; 2004a). However, training and support did not improve attitudes towards working with drinkers and even worsened the attitudes of those who were already insecure and uncommitted (Anderson et al., 2004a). This suggests that training and support should be geared to the needs and attitudes of health professionals to avoid being counterproductive. Anderson et al. (2004b) carried out a meta-analysis of studies testing the effectiveness of different strategies for increasing GPs’ screening and advice-giving rates for hazardous and harmful alcohol consumption. Findings were that, although the paucity of studies suggested caution in interpreting the results, it was possible to increase the engagement of GPs in this activity. While more high-quality research is needed on this topic, promising programmes seemed to be those that had a specific focus on alcohol (rather than general prevention programmes) and those that were multi-component. Part of the problem of translating research into practice in this area is the fact that most trials of brief intervention have been efficacy rather than effectiveness trials (Flay, 1986); that is, they provided a test of screening and brief intervention under optimum research conditions rather than under real-world conditions of routine practice. For this reason, research now needs to focus on ways in which the procedures and materials making up screening and brief intervention programmes can be adapted to meet the needs of routine practice, and the requirements and preferences of both practitioners and service users. Current research is being addressed to these aims: • The English arm of Phase IV of the WHO collaborative project referred to above has carried out a Delphi study (a method designed to reach a consensus among experts on a particular topic) on how that adaptation should proceed (Heather et al., 2004) and focus groups with both health professionals and patients concerning their views on this matter (Hutchings et al., 2006) • An action research project funded by the Tyne and Wear Health Action Zone (HAZ) is currently piloting screening and brief intervention in one general medical practice in each of the five HAZ areas. Various methods of screening, intervention, monitoring and specialist support provision are being tried out with the objective of developing an implementation package that is acceptable to all practices taking part • In the Alcohol Harm Reduction Strategy for England (Prime Minister’s Strategy Unit, 2004) the Government has stated its intention to fund pilot studies of implementing targeted screening and brief alcohol Chapter 7.qxp 17/11/2006 11:00 Page 90 Brief interventions 91 interventions. This will include research in primary healthcare, A&E services, the criminal justice system and possibly other settings. 7.11.3 Conclusions • Most healthcare professionals have yet to incorporate screening and brief interventions for hazardous and harmful drinking into their routine practices (III) • GPs in particular tend to miss most hazardous and harmful drinkers presenting to their practices (I) • Specific barriers to the implementation of screening and alcohol brief interventions in primary healthcare have been identified, including lack of time and lack of suitable reimbursement (I) • Telemarketing appears to be the most cost-effective strategy for disseminating screening and brief intervention packages in primary healthcare (IB) • Training and support can increase the implementation of screening and alcohol brief intervention in primary healthcare (IB) • Training and support should be carefully adapted to meet the needs and attitudes of healthcare professionals (I) • Research should focus on the effectiveness of brief interventions in real world conditions and on ways in which screening and intervention can be successfully implemented in healthcare settings (IV). Chapter 7.qxp 17/11/2006 11:00 Page 91 92 Review of the effectiveness of treatment for alcohol problems Implications for… Service users and carers • These are not interventions that will normally be targeted at help-seekers • Expect more screening and brief interventions for problem lifestyle behaviours in all healthcare settings and other opportunistic points of contact. Service providers • Ensure that protocols and care pathways allow for screening and brief interventions • Build role legitimacy for delivering brief interventions among staff in generic services • Understand the place and limitations of screening and brief interventions • Support training to deliver and incorporate brief interventions into routine practice. Commissioners • Understand the place and limitations of screening and brief interventions – in the main, the evidence is only for generalist settings • Provide training and support for generic staff to deliver brief interventions • Implement across settings where effectiveness has been demonstrated • Commissioning of brief interventions in primary care settings would have a major impact on public health. Researchers • UK research is needed on the longer-term effects of brief interventions • In addition to effects on alcohol consumption, future research should study the effects of brief interventions on alcohol problems, general adjustment and mortality • More UK research is needed to clarify the effects of brief interventions delivered on hospital wards • The effectiveness of brief interventions in several other medical settings requires evaluation, including prenatal and psychiatric services • UK research is urgently needed on the effects of brief interventions in a range of non-medical settings, including social services, the criminal justice system and the workplace • Research is needed to clarify what additional advantages can be expected from extended brief interventions compared to simple, structured advice • Research should also investigate the characteristics of clients who are most likely to respond to simple or to extended brief interventions • A major research effort is required to find ways of implementing and maintaining the delivery of brief interventions in routine practice in a range of medical and non-medical settings and how the barriers to such implementation can be successfully overcome. Chapter 7.qxp 17/11/2006 11:00 Page 92 Less-intensive treatment 95 single session of “advice counselling”, also involving the spouse. At follow-ups six, 12 and 18 months after the initial appointment, both groups showed significant improvements on all marital adjustment and alcohol- related outcome measures, but there were no significant differences between groups. The authors concluded that a single session of advice counselling was as effective as eight sessions but warned that couples in the study represented a socially stable group with a moderate level of alcohol-related difficulties and relatively non-distressed marital relationships. Therefore, this should be the target population for this form of less-intensive treatment. 8.4.3 Conclusion • A single session of conjoint marital therapy is effective among socially stable alcohol misusers with moderate dependence and alcohol problems and relatively intact marriages (IB). 8.5 Motivational interviewing 8.5.1 Context The most popular forms of less-intensive treatment currently available are based on the set of therapeutic principles and counselling techniques known as motivational interviewing (Miller and Rollnick, 1991; 2002). Motivational interviewing is closely linked with the stages of change model described in chapter one. This approach to treatment of alcohol problems fits with the following observations: • Many people who present to agencies for treatment of alcohol problems have not yet formed a definite commitment to change • Even when an alcohol misuser seems convinced that change is necessary, there is often a lingering attachment to heavy drinking and intoxication, and a deep ambivalence towards alcohol • Conflict is an essential part of what we mean by addiction or dependence (Orford, 2001). Motivational interviewing includes a collection of therapeutic principles, a set of counselling techniques and, more generally, a style of interaction. It is defined by Miller and Rollnick (2002, p25) as “a client-centred, directive method for enhancing intrinsic motivation to change by exploring and resolving ambivalence.” The guiding principles of the therapist’s interaction with the service user are: • Express empathy • Develop discrepancy • Roll with resistance • Support self-efficacy. A basic assumption of motivational interviewing, at least as a standalone treatment, is that, once motivated to change, service users can succeed in doing so by using their own change resources and without additional training in behaviour change skills. A full account of the theory, principles and techniques of motivational interviewing is given by Miller and Rollnick (2002). Motivational interviewing is contrasted with the traditional confrontational approach to alcoholism treatment in table 8a. Given the popularity of the confrontational approach, there is surprisingly little evidence to support it. Alcohol misusers at all levels of severity do not show more denial Confrontational approach Motivational approach Heavy emphasis on acceptance of self as “alcoholic”; acceptable of diagnosis seen as essential for change De-emphasis on labels; acceptance of “alcoholism” label seen as unnecessary for change to occur Emphasis on disease of alcoholism which reduces personal choice and control Emphasis on personal choice regarding future use of alcohol and other drugs Therapist presents perceived evidence of alcoholism in an attempt to convince the service user of diagnosis Therapist conducts objective evaluation but focuses on eliciting the service user’s own concerns. Resistance seen as “denial”, a trait characteristic of problem drinkers requiring confrontation Resistance seen as an interpersonal behaviour pattern influenced by the therapist’s behaviour; resistance is met with reflection Table 8a: Differences between confrontational and motivational approaches Chapter 8.qxp 17/11/2006 10:59 Page 95 Review of the effectiveness of treatment for alcohol problems 96 and resistance than people without drinking problems. Those who accept the label of alcoholism do no better, and may actually do worse, than those who reject it (Miller and Rollnick, 1991). When compared to alternative approaches to counselling, confrontation has been found to be less effective in general and to be harmful for service users with low self-esteem (Annis and Chan, 1983). It is important to note here that the confrontational approach runs entirely counter to the spirit of the writings of Bill Wilson, the co-founder of Alcoholics Anonymous (AA World Services, 1980) and to the treatment philosophy underpinning the 12-Step method (see chapter 12). Miller, Benefield and Tonigan (1993) provided strong support for an interactional view of service user motivation. They randomly assigned alcohol misusers to receive confrontational counselling or a client-centred motivational counselling style. Service users in the confrontation group showed much higher level of resistance during counselling sessions than those in the other group. In addition, the more the counsellor had used a confrontational style during counselling, the greater the service user’s alcohol consumption at follow- up over a year later. This and other evidence (Miller and Rollnick, 2002) strongly suggests that confrontation is counterproductive in the attempt to motivate service users for treatment and that a non-confrontational approach should be preferred (see also chapter four). 8.5.2 Evidence The category of motivational enhancement occupies second place in the Mesa Grande (see page 44), although many of the studies included there were of opportunistic brief interventions and were not carried out among treatment samples. Five systematic reviews of research on the effectiveness of motivational interviewing (MI) for a range of addictive disorders have been published. Noonan and Moyers (1997) reviewed 11 clinical trials evaluating MI, nine with alcohol misusers and two with “drug abusers”. Their conclusion was that: “Most of these studies support MI as a useful clinical intervention. MI appears to be an effective, efficient and adaptive therapeutic style worthy of further development, application and research” (p8). Dunn, DeRoo and Rivara (2001) reported a systematic review of MI covering 29 randomised trials over the four behavioural domains of substance abuse, smoking, HIV risk-taking, and diet and exercise. The authors concluded: “There was substantial evidence that MI is an effective substance abuse intervention method when used by clinicians who are non-specialists in substance abuse treatment, particularly when enhancing entry to and engagement in more intensive substance abuse treatment-as-usual” (p1725). Therefore, MI can be used as a preparation for the more intensive forms of treatment discussed in the next chapter (chapter eight). Three systematic reviews of MI have recently been published by Brian L Burke and colleagues. Burke, Arkowitz and Dunn (2002) began by noting that virtually all published research in this area involves the study of adaptations of MI (AMIs), rather than MI in its relatively pure form. AMIs refer to “packaged” versions of MI in which certain methods, such as feedback of assessment results, are used as a shortcut to elicit the service user’s reflections on the pros and cons of the behaviour in question, such as a drinker’s check-up (Miller, Sovereign and Krege,1988), motivational enhancement therapy (Miller et al., 1992) and brief motivational interviewing (Rollnick, Heather and Bell, 1992). The reviewing method used by Burke and colleagues was based on the “box score” method developed by Miller et al. (1995) and, as noted in chapter three, this has been criticised by Finney (2000). However, the earlier review by Burke, Arkowitz and Dunn (2002) was superseded by later work by Burke, Arkowitz and Menchola (2003) that used quantitative meta-analysis in a technically sophisticated manner. None of the conclusions reached by Burke, Arkowitz and Dunn were overturned by this later review. The authors identified 30 controlled trials that met their inclusion criteria, of which 15 were in the area of alcohol problems: • Two trials (Bien, Miller and Boroughs, 1993; Brown and Miller, 1993) looked at AMI as a prelude to treatment among service users at the more severe end of the range of alcohol-related problems. Both found clear evidence of the effectiveness of AMI for this specific purpose • Thirteen trials considered AMI as a standalone intervention. Clear interpretation of research on AMIs as a standalone intervention from this review is difficult, because this category of studies combines the separate domains of Chapter 8.qxp 17/11/2006 10:59 Page 96 Less-intensive treatment 97 opportunistic intervention in the non-treatment–seeking population and less-intensive treatment in the treatment- seeking population. Nevertheless, on balance, the evidence suggested MI-based interventions among a diverse range of groups were effective, including those with significant dependence seeking help for established alcohol problems. Effect sizes were in the small to medium range for comparisons of AMIs with placebo or no treatment conditions. There was no evidence that AMIs were superior to alternative forms of treatment for alcohol problems, but here the MI-based intervention was usually less intensive than the comparison treatment, suggesting that it may be more cost-effective. In the latest review by this team, Burke et al. (2004) updated the conclusions of their previous meta-analysis by including 38 studies of AMI. These conclusions were not substantially changed. The authors also provided answers to other questions regarding AMI: • There was some evidence that MI achieves its effects in the theoretically expected manner by increasing motivation or readiness to change. However, there was no current evidence that this mechanism of change was specific to AMIs as opposed to other forms of intervention • With special regard to AMI as a prelude to other treatment, there was a suggestion that it works by increasing treatment participation, but no firm evidence of a mediating role for increased participation in linking AMI and treatment outcome • There were methodological weaknesses in much of the research reviewed. The greatest threats to internal validity arose from lack of proper treatment specification, insufficient attention to treatment fidelity and the rarity of checks on treatment integrity. Finally, Burke et al. considered relationships between AMI and the other major and well-researched modality in the treatment of addictions, cognitive-behavioural skills training (see chapter eight). They concluded that very little is known about the relative effectiveness of these two forms of treatment, whether they are indicated for different types of service user or whether they could be profitably combined in treatment delivery. The three government-sponsored reviews consulted for this document reached the following conclusions with respect to motivational interviewing: • Among its post-detoxification population of interest, the Scottish review (Slattery et al., 2003) concluded that MI was supported as an effective part of more extensive psychosocial treatment (p5–9) • Based partly on its own meta-analysis, the Swedish review (Berglund, Thelander and Jonsson, 2003) concluded that “motivational interviewing increases the effect of another treatment, but has not itself been subjected to randomised study” (p56) • The Australian review (Shand et al., 2003) concluded that: “The effectiveness of motivational interviewing delivered prior to treatment is unclear and there is a need for further studies to address this issue” (p50). The difference in conclusions between the Swedish and Scottish reviews, and the Australian review is that the Australian work highlighted the short, three-month follow- ups on which the favourable findings of the two studies of MI as a prelude to treatment proper (Bien, Miller and Boroughs, 1993; Brown and Miller, 1993) were based. Therefore, several important questions remain regarding the effective mechanisms of MI (and MET – see section 8.6), the duration of its effects and its possible advantages and disadvantages compared to other forms of treatment. However, the relative brevity and cost- effectiveness of MI, combined with its growing popularity among treatment professionals, suggests that it should occupy a prominent place in modern treatment services. 8.5.3 Conclusions • The non-confrontational principles and style of MI should inform the conduct of specialist treatments for alcohol problems (IB) • MI increases the effectiveness of more extensive psychosocial treatment (IA) • While there is no evidence at present of long-term effects, MI and its adaptations can be effective as a preparation for more intensive treatment of different kinds (IA) • Standalone adaptations of MI are no more effective than other forms of psychosocial treatment but are usually less intensive and therefore potentially more cost-effective (IA). Chapter 8.qxp 17/11/2006 10:59 Page 97